SteeraMed Core

English | 中文

SteeraMed: A Steerable Biomedical World Model — Personalized intervention evidence chains from DNA methylation data for longevity, aging, and chronic diseases.

Select a patient case → Generate individualized drug evidence in 30 seconds. SteeraMed.com · Paper

What is a Biomedical World Model?

Traditional systems biology:

• Population statistics → average effects → universal guidelines
• "Is this drug effective for the population?"

Steerable Biomedical World Model (SBWM):

• Individual perturbation → matched PPI modules → personalized evidence chain
• "Is this drug effective for you?"

Key differences:

	Systems Biology	Steerable Biomedical World Model
Unit of analysis	Population	Individual (N-of-1)
Question	Group average	Personal match
Output	General guideline	4-layer evidence chain
Drug ranking	Clinical trials	SA alignment + bootstrap

Four-Layer Evidence Chain

Layer 1: PPI Module Perturbation  ← "What's different in your biology?"
Layer 2: Compound SA Alignment    ← "Which compounds can correct it?"
Layer 3: Mechanism Annotation     ← "Why does this compound work?"
Layer 4: Bootstrap Confidence     ← "How reliable is this result?"

Quick Start

pip install steeramed-core
python -m steeramed_core

Interactive case selector:

🧬 SteeraMed Core — N-of-1 Evidence Chain Explorer
═════════════════════════════════════════════════════

Select a patient case:
  [1] 🧓 Aging · Population Screening
  [2] 🧑 RA · 51M · T-cell Perturbation
  [3] 🧑 Depression · 52M · Innate Immunity

Enter choice [1-3]: 2

✅ Generated 4 figures in results/:
  📊 hallmark_bar.png      — Hallmark perturbation profile
  💊 drug_ranking.png       — Top-10 compound ranking
  🔗 evidence_network.png  — Drug-PPI-Hallmark alignment
  📋 patient_card.png       — One-page patient summary

Batch mode:

python -m steeramed_core --all          # all cases
python -m steeramed_core --case ra_303  # specific case
python -m steeramed_core --list         # list available cases

Patient View Gallery

Aging Patient View — Population-level screening (GSE40279, N=473 old vs young). Three-panel card showing perturbed aging hallmarks, top-10 compound ranking (Niacin #1), and bootstrap confidence:

Depression Patient View — N-of-1 case: 52-year-old male (GSE128235). Innate immunity–dominant perturbation profile with creatine as top-ranked compound:

More figures

Scientist View — RA Evidence Chain (GSE42861, 51M):

Patient View — RA (GSE42861, 51M):

Scientist View — Depression Evidence Chain (GSE128235, 52M):

Scientist View — Aging Evidence Chain (GSE40279):

Hallmark Perturbation Bar (Aging):

Available Cases

Case	Disease	Key Finding	Evidence
Aging · Population	GSE40279	Niacin #1, 2/5 geroprotectors	MODERATE
RA · 51M	GSE42861	6/10 known RA drugs, pentoxifylline #1	STRONG
Depression · 52M	GSE128235	creatine #1, innate immunity	EXPLORATORY

API

import json
from pathlib import Path
from steeramed_core.viz.patient_card import plot_patient_card
from steeramed_core.viz.drug_ranking import plot_drug_ranking

p = Path(__file__).parent / "steeramed_core" / "presets" / "example_patients"
data = json.loads((p / "ra_patient_303.json").read_text(encoding="utf-8"))
fig = plot_patient_card(data)
fig.savefig("my_patient_card.png", dpi=300)

Data Sources & Acknowledgments

This package includes pre-computed results derived from the following open databases. We gratefully acknowledge the original data providers:

• PPI Network: STRING v12.5 — Szklarczyk et al., Nucleic Acids Res 53(D1), 2025. CC BY 4.0
• Compound–Target Interactions: STITCH — Kuhn et al., Nucleic Acids Res 36(Database), 2008. CC BY-NC — this package uses STITCH-derived data for academic research only; commercial applications require separate authorization from EMBL
• Methylation Data: GEO (NCBI) — public domain
• Hallmark Gene Sets: MSigDB — Liberzon et al., PNAS 112(25), 2015. CC BY 4.0

Note: This repository distributes pre-computed analytical results (e.g., ranked compound lists, PPI module summaries), not the original STRING or STITCH databases. Users who wish to access or redistribute the underlying databases must comply with their respective license terms.

Citation

If you use SteeraMed Core in your research, please cite both companion papers:

• Framework paper: Xiong J. World Models for Biomedicine: A Steerability Framework. doi:10.20944/preprints202605.0366.v1
• Implementation paper: Xiong J. SteeraMed: A Biomedical World Model for N-of-1 Intervention Reasoning across Chronic Diseases and Aging. doi:10.20944/preprints202605.1578.v1

@article{xiong2026steeramed,
  title={SteeraMed: A Biomedical World Model for N-of-1 Intervention Reasoning across Chronic Diseases and Aging},
  author={Xiong, Jianghui},
  journal={Preprints.org},
  year={2026},
  doi={10.20944/preprints202605.1578.v1}
}
@article{xiong2026framework,
  title={World Models for Biomedicine: A Steerability Framework},
  author={Xiong, Jianghui},
  journal={Preprints.org},
  year={2026},
  doi={10.20944/preprints202605.0366.v1}
}

Disclaimer

This software generates hypothesis-generating insights only. It is not a medical device and does not provide treatment recommendations. Always consult qualified healthcare professionals for medical decisions.

Keywords

biomedical world model · medical world model · steerability · longevity · aging · personalized medicine · n-of-1 · DNA methylation · epigenetics · drug ranking · PPI network · evidence chain · precision medicine · intervention reasoning · rheumatoid arthritis · depression · hallmark · methylation age

License

MIT License. This applies to the code in this repository only.

The pre-computed data files (JSON files in steeramed_core/presets/) incorporate derivative results from STITCH (CC BY-NC). These data files are provided for academic research and educational purposes only. Commercial use of STITCH-derived data requires authorization from EMBL.