Robust and optimal predictive control of the COVID-19 outbreak

Area of Science:

• Epidemiology
• Control Theory
• Public Health Policy

Background:

• The COVID-19 pandemic necessitates effective control strategies to minimize fatalities and social costs.
• Optimal control policies are crucial for managing infectious disease outbreaks like COVID-19.
• Germany serves as a case study for implementing and evaluating pandemic control measures.

Purpose of the Study:

• To investigate adaptive strategies for robust and optimal control of the COVID-19 pandemic using social distancing.
• To minimize fatalities over a two-year period while avoiding excessive social costs.
• To design and validate control approaches using a tailored model of the German COVID-19 outbreak.

Main Methods:

• Utilized an open-loop optimal control policy for comparison under exact model knowledge.
• Implemented a weekly updating feedback strategy using model predictive control (MPC) for uncertain scenarios.
• Developed a robust MPC-based feedback policy incorporating interval arithmetic for cautious adaptation.

Main Results:

• Open-loop control significantly reduces fatalities compared to simpler policies with exact model knowledge.
• MPC feedback strategy demonstrates reliable performance even with model mismatch and deviant parameters.
• Robust MPC policy minimizes fatalities with inaccurate measurements and unspecified infection rates.

Conclusions:

• Adaptive feedback strategies are essential for reliably containing the COVID-19 outbreak.
• Well-designed policies can substantially decrease fatalities without increasing social distancing measures.
• Early, stronger social distancing is more effective and cost-efficient than delayed, fluctuating measures.