digitalsoma 2.2.0

A physics-based digital twin framework for real-time animal physiology monitoring

DigitalSoma

A physics-based digital twin framework for real-time animal physiology monitoring

Author: Dr. ir. Ali Youssef — Adjunct Professor, Computational Bio-Ecosystems, Agroecosystems Laboratory, University of Manitoba & BioTwinR Ltd., Winnipeg, Canada

---

Install

pip install digitalsoma

# Optional extras
pip install "digitalsoma[yaml]"   # PyYAML support
pip install "digitalsoma[llm]"    # Anthropic SDK for LLM agent interface
pip install "digitalsoma[dev]"    # pytest + build tools

---

60-second quick start

from digitalsoma import build_soma, SomaConfig

# Build a porcine digital twin
ds = build_soma(SomaConfig(animal_type="porcine_adult", animal_id="pig-001"))

# Ingest sensor readings — vendor aliases resolved automatically
state = ds.update_sync({
    "HR":    110,    # alias → heart_rate_bpm
    "Tb":    40.1,   # alias → core_temp_C
    "SpO2":  97.2,   # alias → spo2_pct
    "RR":    52,     # alias → respiratory_rate_bpm
    "cort":  80.0,   # alias → cortisol_nmol_L
})

# Six solvers run automatically
print(state["cardiac_output_L_min"])        # Fick: 6.6 L/min
print(state["thermal_comfort_index"])        # Newton: 0.94 → heat stress
print(state["physiological_stress_index"])  # HPA: 0.52
print(state["adverse_event_score"])         # VeDDRA: 0.17

# FHIR R4 export (new in v2.2.0)
bundle = ds.to_fhir_bundle()
print(bundle["total"])                       # 21 resources

# VeDDRA adverse event report
report = ds.veddra_report()
print(report["clinical_signs"])              # [{veddra_id, veddra_term, ...}]

---

What it does

DigitalSoma represents any living animal as a continuously updated computational object. Raw sensor readings (heart rate, temperature, accelerometry, blood markers) enter through a schema-agnostic manifest layer, are normalised against internationally recognised ontology vocabularies, and pass through a composable chain of physics-based physiological solvers that infer clinically meaningful state variables in real time.

Five-layer architecture

Input sources
  Wearable · Implanted · Remote sensing · Lab assay · Manual entry
          │
          ▼
Ontology & Normalisation Layer          (digitalsoma/ontology/vocab.py)
  Uberon · SNOMED CT · VeDDRA · NCBITaxon · UCUM · PATO · HP/MP
  canonical_key()  normalise_dict()  to_jsonld()
          │
          ▼
┌─────────────────────────────────────────────────────┐
│  DigitalSoma core  (digitalsoma/soma_api.py)        │
│                                                     │
│  Structural Layer   Dynamic Layer   Functional      │
│  ATR · anatomy ·   KV store O(1)   Model Zoo DAG   │
│  normal ranges     TSL · TES        6 built-in      │
│  build_soma()      update_sync()    solvers          │
└─────────────────────────────────────────────────────┘
          │
          ▼
LLM Agentic Interface Layer             (digitalsoma/soma_agent.py)
  11 OpenAI-compatible tool schemas · SomaDispatcher
          │
          ▼
Outputs
  State snapshot · Time-series · JSON-LD · VeDDRA AE report
  FHIR R4 Bundle · LLM response

Six built-in solvers (DAG execution order)

#	Solver	Model	Key outputs
S1	cardiovascular_baseline	Fick equation	cardiac_output_L_min
S2	metabolic_rate	Kleiber's law + Q10	rmr_W, rmr_kcal_day
S3	thermoregulation	Newton's cooling law	thermal_comfort_index
S4	respiratory_gas_exchange	Respiratory quotient	vo2_L_min, minute_ventilation_L_min
S5	neuroendocrine_stress	HPA axis composite	physiological_stress_index
S6	adverse_event_screen	VeDDRA v2.2 mapping	ae_flags, adverse_event_score

Custom solvers plug in via ds.register_method(name, fn) and slot into the same DAG.

Animal Type Registry — 6 species templates

Key	Species	Body mass	HR baseline
porcine_adult	Sus scrofa domesticus	90 kg	75 bpm
equine_adult	Equus caballus	500 kg	36 bpm
bovine_adult	Bos taurus	600 kg	60 bpm
ovine_adult	Ovis aries	70 kg	75 bpm
canine_adult	Canis lupus familiaris	25 kg	90 bpm
salmonid_adult	Salmo salar	4.5 kg	50 bpm

Custom species registered via register_animal_type(name, config).

---

HL7 FHIR R4 integration

New in v2.2.0 — zero external dependencies:

from digitalsoma.fhir import to_fhir_bundle, from_fhir_bundle

# Export: Patient + Device + Observations (LOINC + SNOMED CT + UCUM) + DiagnosticReport
bundle = ds.to_fhir_bundle()                        # collection bundle
tx     = ds.to_fhir_bundle(bundle_type="transaction") # POST to FHIR server

# Ingest: parse incoming FHIR Observations → readings dict → update_sync()
readings = from_fhir_bundle(incoming_bundle)
ds.update_sync(readings)

VeDDRA adverse event findings appear in the DiagnosticReport dual-coded with SNOMED CT and VeDDRA term IDs — ready for EMA EVVET3, UK VMD, and FDA-CVM submission.

---

VeDDRA pharmacovigilance

Six clinical signs screened on every update_sync() call:

VeDDRA term	VeDDRA ID	Trigger	SNOMED CT
Hyperthermia	10020557	T > T_base + offset	386689009
Hypothermia	10021113	T < T_base − offset	386692006
Tachycardia	10043071	HR > HR_base × 1.5	3424008
Bradycardia	10006093	HR < HR_base × 0.6	48867003
Hypoxia	10021143	SpO₂ < 90 %	389086002
Distress	10013029	PSI > 0.70	274668005

---

Ontology namespaces

Namespace	URI base
Uberon	http://purl.obolibrary.org/obo/UBERON_
SNOMED CT	http://snomed.info/id/
VeDDRA	https://www.ema.europa.eu/en/veterinary-regulatory/
NCBITaxon	http://purl.obolibrary.org/obo/NCBITaxon_
UCUM	http://unitsofmeasure.org/
LOINC	http://loinc.org

---

Examples

python -m digitalsoma.examples.e1   # five species, structural layer
python -m digitalsoma.examples.e2   # heat-stress / recovery cycle
python -m digitalsoma.examples.e3   # ontology compliance, alias resolution
python -m digitalsoma.examples.e4   # custom HRV solver + VeDDRA report
python -m digitalsoma.examples.e5   # FHIR R4 export and round-trip

Or run directly from the source tree:

python examples/e5_fhir_integration.py

---

Tests

pip install "digitalsoma[dev]"
pytest tests/ -v

---

Citation

@software{youssef2026digitalsoma,
  author    = {Youssef, Ali},
  title     = {{DigitalSoma}: A physics-based digital twin framework
               for real-time animal physiology monitoring},
  year      = {2026},
  version   = {2.2.0},
  publisher = {BioTwinR Ltd. \& University of Manitoba},
  url       = {https://github.com/Pierianspring/digitalsoma},
  orcid     = {0000-0002-9986-5324},
}

See CITATION.cff for the CFF-format citation (GitHub "Cite this repository" button).

---

Licence

CC BY 4.0 — see LICENSE. Free to use, share, and adapt with attribution.

---

Links

• PyPI: https://pypi.org/project/digitalsoma/
• GitHub: https://github.com/Pierianspring/digitalsoma
• Documentation: https://github.com/Pierianspring/digitalsoma/tree/main/docs
• ORCID: https://orcid.org/0000-0002-9986-5324
• Issues: https://github.com/Pierianspring/digitalsoma/issues