Products

Clinical AI · Knowledge Graph · Multi-Agent

Health

A biomedical knowledge graph workspace for clinical reasoning. 270K+ nodes, 12M+ relationships, 8 clinical roles, and vision-grounded multi-agent consults, all safety-gated.

270K+
KG Nodes
12M+
Relationships
8
Clinical Roles
82%
Hits@10
View on GitHub ↗
PatientDiseaseDrugGeneProteinSymptom

Knowledge Sources

3 graphs. Unified.

PrimeKG, DRKG, and HetioNet merged into a single Neo4j graph. RotatE embeddings trained for link prediction. 82% Hits@10 on drug-disease associations.

PrimeKGPrecision medicine knowledge graph: diseases, drugs, genes, proteins
DRKGDrug Repurposing Knowledge Graph. 12M+ drug-disease-gene relationships
HetioNetHeterogeneous biological network: diseases, compounds, pathways

Role-Based Workflows

8 clinical roles. One platform.

Each role gets a tailored dashboard with unique metrics, worklist, and action buttons. The system detects the role and adapts the entire UI.

Chief Doctor
Full patient oversight, multi-agent consultation
ED Physician
Emergency triage with high-priority queue
Nurse Manager
Care coordination and medication tracking
Front Desk
Patient intake and appointment management
Hospital Admin
Workforce and compliance dashboards
System Admin
User management and system health
Specialist
Domain-specific consult workflows
Radiologist
Vision-grounded imaging analysis

Clinical Encounter

6-step workflow. Safety-gated.

Progressive disclosure through intake, imaging, transcription, memory, multi-agent reasoning, and clinician review, with validation gates at every step.

01
Patient
Identity, demographics, existing conditions loaded from record
02
Capture
Image acquisition: radiology, dermatology, retinal scans
03
Transcript
Real-time transcription of clinician + patient conversation
04
Memory
Longitudinal graph: prior encounters, labs, external records
05
Reasoning
KG-grounded multi-agent consultation with confidence scoring
06
Review
Clinician decision with safety gates and audit trail

Multi-Agent Consult

Agents with confidence.

Each specialist agent returns a confidence score, evidence list, and recommendation. Clinicians see all agents side-by-side before making a decision.

Radiology Vision Agent91% confidence
Pulmonology Agent87% confidence
Medication Safety Agent94% confidence
Oncology Agent78% confidence

Knowledge-grounded clinical AI.

Reducing hallucinations through 12M+ biomedical relationships.

GitHub ↗