AI model predicts TB outbreak in Mars colonies using PPO-based control
Stochastic simulation reveals latent TB can reactivate even without initial cases.
A new study by Teddy Lazebnik introduces a stochastic host-radiation-pathogen-habitat model for latent tuberculosis reactivation in a Mars colony. The model accounts for galactic cosmic radiation degrading immune competence, which can trigger reactivation of latent TB in a closed, small population. Countermeasure allocation is framed as a partially observable sequential decision problem, with a proximal policy optimization (PPO) agent trained on an agent-based simulator to choose isolation and medication strategies.
Simulations show that even without any initial infectious cases, active TB can emerge endogenously. The risk is most sensitive to the size of the latent reservoir, the strength of radiation-immune coupling, and reactivation sensitivity. Adaptive control using PPO significantly reduced infectious burden and mortality while cutting unnecessary interventions. The framework can be used for pre-launch stress-testing of screening, monitoring, shielding, and treatment strategies tailored to specific mission parameters.
- PPO-trained agent optimizes isolation and medication allocation in a closed Mars habitat.
- Endogenous TB outbreaks occur even with zero initial infectious cases due to radiation-driven immune decline.
- Risk hinges on latent reservoir size, radiation-immune coupling, and reactivation sensitivity.
Why It Matters
AI-driven simulations enable pre-mission planning for infectious disease containment in isolated space colonies.