Scientific Architecture

The Knowledge Graph™
Behind CardioIQ

Cardiovascular medicine, organized as a connected semantic network. Every recommendation traceable. Every inference evidence-anchored. Every risk visible.

Get my report — from $29 → Download White Paper Explore the architecture →
7 Semantic Layers
14 Relationship Types
6 Guideline Bodies
0.99 Max Confidence Score
Diagram 1

Architecture Overview

Patient data flows through a structured eight-stage pipeline. The Knowledge Graph sits at the centre, connecting all clinical reasoning layers into a single explainable output.

Patient Data
Labs · History · Comorbidities · Medications
Clinical Normalization Layer
Unit conversion · Terminology standardization
Knowledge Graph™ (7 Layers)
Biomedical · Disease · Risk · Therapeutic · Guideline · Evidence · Patient
Reasoning Engine
Deterministic · Explainable · Auditable
Evidence Engine  ·  Guideline Engine
PMID/DOI anchoring · Class & Level mapping
Clinical Intelligence Engine
Risk stratification · Gap detection · Recommendations
Clinical Report
Patient summary · Clinician decision support

Every recommendation in the output is traceable back through this pipeline to a specific patient finding and guideline statement.

Knowledge Graph Structure

Seven Semantic Layers

The Knowledge Graph is not a single flat database. It employs seven interconnected layers — each independently expandable as the evidence base evolves.

Layer 1

Biomedical

Genes · Proteins · Biomarkers · Metabolic Pathways

Layer 2

Disease

ASCVD · FH · Hypertension · Diabetes · CKD · NAFLD · Obesity · HF · PAD · Stroke

Layer 3

Risk

Absolute Risk · Lifetime Risk · Residual Risk · Inflammatory Risk

Layer 4

Therapeutic

Lifestyle · Medication · Monitoring · Follow-up Protocol

Layer 5

Guideline

ESC/EAS · ACC/AHA · ADA · KDIGO · USPSTF · IAS

Layer 6

Evidence

RCTs · Meta-analyses · Consensus Statements · Registries · PMID/DOI

Layer 7

Patient

Individual patient subgraph — generated dynamically for each assessment, creating a personalised risk representation.

Semantic Relationships

How the Graph Reasons

Every connection in the Knowledge Graph carries a semantic type, an evidence level, and a confidence score. This is what makes CardioIQ's reasoning auditable — not a black box.

CAUSES
ApoB → Atherosclerosis
Class I, Level A · Confidence 0.99
AMPLIFIES
CKD → ASCVD Risk
KDIGO 2022
INCREASES
Lp(a) → Residual Risk
EAS Consensus 2022
ACCELERATES
Diabetes → Atherosclerosis
ESC 2023, Class I
TREATS
PCSK9i → Elevated ApoB
Class I, Level A
MODIFIES
Residual Risk → Treatment Strategy
ESC/EAS 2019

All 14 supported types: CAUSES · ASSOCIATED_WITH · PREDICTS · RECOMMENDS · TREATS · CONTRAINDICATED_IN · REQUIRES · MODIFIES · AMPLIFIES · SUPPRESSES · INTERACTS_WITH · DERIVED_FROM · SUPPORTED_BY · SUPERSEDES

Design Principles

Built on Five Core Principles

CardioIQ Knowledge Graph™ is architected to be trustworthy by design — not by claim.

Biological Accuracy

Every relationship reflects accepted human pathophysiology and established biochemical mechanisms.

Clinical Validity

Every edge is supported by peer-reviewed evidence or internationally recognised guideline recommendations.

Explainability

Every inference is traceable and human-understandable — decomposable into patient findings, graph paths, and source evidence.

Modularity

Domains can be expanded independently — lipids, inflammation, renal, metabolic, vascular ageing — without structural redesign.

Version Control

Every node and relationship includes creation date, version number, source reference, evidence level, guideline version, and confidence score.

Scientific White Paper · Version 1.0
CardioIQ Knowledge Graph™
Full Technical Specification
The complete architecture document — intended for clinicians, data scientists, healthcare innovators, and partners exploring evidence-based cardiovascular AI.
Tea Gamezardashvili, MD, PhD, MHA, FACC
14 pages
1.0 · July 2026
PDF
Download White Paper (PDF)
CIQ
CardioIQ Knowledge Graph™
White Paper · V1.0

See the Knowledge Graph applied to your labs

Enter the values from your existing panel and a lipidologist reads them against longevity-optimal targets — a structured report in your inbox within 2 business days.

Get my report — from $29 →