name the institution, drop its files, then "Add" · repeat for more · at least Baseline required
Loaded institutions
Learning Epidemics
Social Network Analysis Dashboard
Institution
No data loaded
↑ Load Data
Baseline 2D
Baseline 3D
Midline
Endline
Longitudinal
Network Evolution
Learning Epidemic
Descriptive Stats
Correlation
Random Forest
Simulation Sandbox
Display
Labels
Arrows
Target-only nodes
Weighted edges
Node size by
Color nodes by
Edge Filters
Min weight
0
Max rank shown
10
Node Filters (source)
Sex
Department
Role
Node Isolation
Search node (ego network)
Neighbourhood depth
1-hop
Show only truly isolated nodes
Dim non-ego nodes
Color Coding
Sex
Dept
Role
Legend
University of Arizona
Mel & Enid Zuckerman College of Public Health in partnership with Teya
PI Onicio B. Leal Neto Co-PI Alexandre Santilli Team Julia Nicolau
Node Metrics
Edge List
Load a CSV to see data
Network Metrics
Nodes
—
Edges
—
Density
—
Components
—
Avg Degree
—
Reciprocity
—
Clustering
—
Avg Path
—
Degree Assortativity (Newman's r)
—
Assort. Sex
—
Assort. Dept
—
Assort. Role
—
Top Nodes
Rank by
Selected Node
3D Display
Node labels
Show edges
Auto-rotate
Target-only nodes
Layout
Node size by
Color nodes by
Force Simulation
Repulsion
10
Link distance
60
Link strength
0.5
Camera
🖱 Drag — orbit
⚙ Scroll — zoom
↗ Right-drag — pan
Depth fog
30
Legend
3D Force-Directed Network Three.js · WebGL
Descriptive Statistics
Summary of all variables — source nodes only
Load a CSV to see statistics
Correlation Matrix
Pearson correlations among numeric variables · source nodes only
Load a CSV to see correlations
Random Forest — Feature Importance
Variable importance for predicting centrality · 100 bootstrap trees · MSE variance reduction
Load a CSV to run analysis
Simulation Sandbox
What-if scenarios on the current institution's network topology & R₀ · 70/30 train–test + synthetic externality testing
1 · Turnover
2 · Key Person / Position
3 · Successor Prep
4 · Change Impact
Every run fits on a 70% training split and validates on a 30% hold-out. The synthetic set generates new employees whose attributes are drawn 70% from this institution and 30% from the other loaded institutions (externality), then attaches them to the network by preferential attachment. New-employee count is taken from the numeric field above.
⬡ Combined Network — connections color-coded by wave of first appearance
● Baseline● New T1● New T2● Shared B+M● All waves
Network Evolution
How key metrics changed across data collection waves · intervention points marked
Metric trajectories over time
● Baseline● Midline● Endline▲ Intervention
Persistent Nodes — present across all loaded waves
● Baseline● Midline● Endline
Learning Epidemic Simulator
What would happen if you launched an AI training program in this organization today?
💉
Step 1 — Seed
Choose how many people start the program (I₀) and from which network structure (Baseline, Midline, or Endline).
🔁
Step 2 — Spread
Adjust β (how contagious the topic is — how often contacts discuss it) and γ (how fast people go from hearing about it to actually adopting it).
📉
Step 3 — Saturation
Set δ (how fast active practitioners stop spreading — through saturation, forgetting, or loss of interest).
📊
Step 4 — Read R₀
R₀ = β/γ tells you how self-sustaining the program is. R₀ > 1 means it spreads on its own. R₀ < 1 means it needs constant re-injection.
Network-informed simulation: the department R₀ values below are calculated directly from the actual communication patterns you measured.
Departments with denser within-group ties get a higher local β, meaning the topic spreads faster internally.
The overall N comes from the real number of survey participants in the selected wave.
Think of it as: if you drop a learning "pathogen" into this exact organizational network, how does it travel?
Simulation Parameters
β — How often contacts share the knowledge
0.30
Low → people rarely discuss it · High → frequent conversations
γ — How fast exposure turns into adoption
0.20
Low → long latency · High → quick uptake
δ — How fast active adopters stop spreading
0.10
Low → sustained diffusion · High → rapid burnout
I₀ — People who start the program (seeds)
3
The initial cohort of the institutional program
Time horizon (days)
120
Network to simulate
Organizational R₀
—
R₀ = β ÷ γ · average people one adopter reaches
Simulation Outcome
Model Equations
dS/dt = −β·S·I/N
dE/dt = +β·S·I/N − γ·E
dI/dt = +γ·E − δ·I
dR/dt = +δ·I
Integrated via Runge-Kutta 4 (dt=0.5 days)
Learning Diffusion Curves
● S Unaware● E Exposed● I Actively spreading● R Knowledge retainedsolid = model · ┄ dotted = live network
Department-Level R₀
derived from actual within-department communication density
Live Network Diffusion — watch the learning spread node-by-node