Reinforcement learning relieves the vaccination dilemma.

Area of Science:

• Epidemiology
• Artificial Intelligence
• Network Science

Background:

• Epidemic spreading is a significant public health concern.
• Decision-making rules influence vaccination behavior and coverage.
• Complex networks present challenges for disease transmission modeling.

Purpose of the Study:

• To investigate how a decision-making rule, based on the Bush-Mosteller (BM) model from artificial intelligence (AI), impacts epidemic spreading on complex networks.
• To analyze the effect of two independent rules (fixed loss and average payoff) on vaccination behavior.
• To determine if AI-driven learning can help overcome the vaccination dilemma.

Main Methods:

• Utilized the Bush-Mosteller (BM) model, a reinforcement learning methodology, incorporating vaccination and epidemiological processes.
• Modeled agent vaccination behavior updates based on fixed loss consideration and average payoff of neighbors.
• Simulated scenarios with varying stimuli, including loss of payoffs and environmental changes.

Main Results:

• Higher sensitivity in decision-making rules led to increased vaccination coverage rates.
• Increased sensitivity delayed the transition point in relative vaccination costs from full to incomplete vaccination.
• The vaccination dilemma was overcome to some extent, with vaccination probabilities showing normal or skewed distributions.

Conclusions:

• The Bush-Mosteller (BM) model, integrated with AI, offers a promising approach to enhance vaccination strategies.
• Optimized decision-making rules can significantly improve vaccination coverage and mitigate epidemic spread.
• AI-empowered learning has the potential to resolve the persistent vaccination dilemma in public health.