Using the American Heart Association's knowledge graph to demonstrate each visualization technique
The American Heart Association's strategic landscape, analyzed through InfraNodus text network analysis, reveals 7 topic clusters with 95 nodes and a critical structural gap between AI & Digital Medicine and Behavior Change.
This guide walks through each visualization in the viz-playground suite using this real-world dataset, showing how different rendering techniques reveal different strategic insights.
| Cluster | Nodes | Betweenness |
|---|---|---|
|
Health Insights
|
28 | 47 |
|
AI & Digital Medicine
|
18 | 16 |
|
Research Funding
|
16 | 15 |
|
Fundraising Impact
|
12 | 6 |
|
Health Equity & Education
|
10 | 6 |
|
Maternal Health
|
6 | 5 |
|
Behavior Change
|
5 | 5 |
Health Insights sits at the center with a betweenness score of 47 — the core knowledge gateway connecting all strategic domains. At the periphery, Behavior Change (5 nodes, betweenness 5) is the smallest and most isolated cluster, with no direct connection to AI & Digital Medicine.
The constellation effect renders all edges at opacity 0.04–0.15 on a near-black background. For AHA's network, this reveals the overall topology at a glance: Health Insights (C1) sits at the center with strong connections radiating outward, while Behavior Change (C7) and Fundraising Impact (C4) occupy the periphery with visibly fainter connections.
The narrow opacity band is critical — it prevents the strong C1↔C3 connection (weight 8) from visually overwhelming the weak C4↔C5 connection (weight 2), while still making the hierarchy perceivable. The entire network reads as a single constellation rather than disconnected fragments.
Key insight: The ABSENCE of an edge between C2 and C7 is visible as empty space — the structural gap literally appears as a gap in the constellation.
Interactive — pan, zoom, and hover to explore
In AHA's network, gap-bridging keywords like "digital therapeutics" or "AI-powered behavior coaching" would receive the full three-tier glow — they connect the disconnected AI and Behavior Change clusters.
The three tiers serve AHA's decision-makers differently:
Hub nodes like "cardiovascular research" (C1, betweenness=47) get a subtler two-tier glow — they're important but expected. The gap nodes are the strategic surprises.
Interactive — pan, zoom, and hover to explore
Force parameters reveal different aspects of AHA's strategic structure:
"Tight Clusters" preset: Shows AHA's organizational silos. Health Insights, Research Funding, and AI Medicine cluster tightly (high mutual connectivity), while Behavior Change drifts to the edge — visually confirming its isolation.
"Spread Exploration" preset: Spreads everything out, making ALL connections visible. Here you can trace the path from Behavior Change to AI Medicine: C7→C5→C1→C2. It exists, but it's indirect — requiring three hops through Health Equity and Health Insights.
"Focus Mode" preset: Pulls everything toward center, compressing the graph. This is the "executive summary" view — the major clusters are visible, and the gap between C2 and C7 is most obvious because they're pulled close but have no connecting edge.
The insight: force parameters aren't just cosmetic — they're analytical tools. Each configuration reveals different structural properties of the same network.
Interactive — adjust sliders and try the presets
With all 19 channels available, AHA's network can encode:
But activating all channels simultaneously overwhelms cognition. The sweet spot for AHA board presentations: Size + Color + Edge Opacity + Glow — four channels that answer "What are the main topics (color), how important are they (size), how connected are they (edges), and where are the strategic gaps (glow)?"
Adding a fifth channel (edge dash for temporal) pushes past cognitive bandwidth — save it for the deep-dive with the research team.
Interactive — toggle channels on/off, watch the cognitive bandwidth indicator
The InfraNodus adapter transforms raw graph exports into the visualization format. For AHA's network, the adapter automatically:
The paste-your-own-data feature means any InfraNodus analysis — competitor landscapes, content audits, research paper networks — can be instantly visualized with the same constellation aesthetic and gap detection.
Interactive — paste your own InfraNodus JSON export to render a custom graph
The artifact template lets Claude generate interactive graph visualizations directly in conversation. Ask Claude to "analyze AHA's strategic landscape and render the knowledge graph" — it fills the data injection point with the analysis results, and the artifact appears as an interactive panel.
Constraints: single file, inline JS/CSS, D3 from whitelisted CDN only. No external API calls, no localStorage. The data lives entirely within the artifact.
Interactive — this is the same output Claude generates as an artifact
The MCP App server brings this system into Claude Desktop. During a conversation, Claude calls render_graph with AHA's network data, and an interactive panel appears alongside the chat.
The bidirectional architecture means:
This closes the loop between analysis and visualization — the agent thinks in graphs, the human sees in graphs, and both can act on what they see.
The single most important finding in AHA's knowledge graph: AI & Digital Medicine (C2) and Behavior Change (C7) have no direct connection. Digital health tools aren't informed by behavioral science, and behavioral interventions aren't leveraging AI.
Every visualization technique in this suite reveals this gap differently:
Strategic Recommendation
Invest in "digital therapeutics" and "AI-powered behavior coaching" — concepts that bridge both clusters and fill the structural gap. These are not incremental improvements. They represent the highest-leverage opportunities in AHA's strategic landscape, sitting at the intersection of two domains that currently operate in isolation.