Stages of Change: Health Behavior Simulator

Population & Stages

Model Parameters

Self-efficacy growth rate γ 0.35

Intervention strength I (support / MI) 0.40

Relapse risk ρ 0.18

Simulation speed 1.0×

Live Readouts

Avg. self-efficacy

—

In Action+Maintenance

—

Cumulative relapses

How it works

Each of the 180 agents carries a self-efficacy value SE ∈ [0,1] (Bandura) and sits in one of Prochaska & DiClemente's five stages of change. Every simulated second, two continuous-time Poisson processes compete:

dSE/dt = γ · (1 − SE) · (0.15 + I)

Decisional balance:  DB = clamp(0.6·I + 0.4·SE − 0.25, 0, 1)
Forward-stage rate:  λ_fwd = 0.22 · SE · DB
Relapse rate (Action/Maintenance only):
                     λ_relapse = ρ · (1 − SE)

P(transition within dt) = 1 − e^(−λ·dt)

Self-efficacy growth γ — how fast confidence in one's own ability to sustain the change builds (CBT, mastery experiences, coaching).
Intervention strength I — motivational interviewing, reminders, social support; raises both the decisional balance (more perceived pros vs. cons) and the SE growth ceiling.
Relapse risk ρ — baseline chance of slipping from Action or Maintenance back to Contemplation; low self-efficacy amplifies it, matching the model's emphasis on relapse as a normal loop rather than a failure endpoint.
Agents never skip a stage forward, but a relapse always returns them to Contemplation — the TTM's "spiral" rather than linear path.

This is the same five-stage model (Precontemplation → Contemplation → Preparation → Action → Maintenance) used in real smoking-cessation, exercise-adoption and medication-adherence interventions.

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.