Fundamental limitations on efficiently forecasting certain epidemic measures in network models.

Area of Science:

• Epidemiology
• Computational Complexity
• Mathematical Biology

Background:

• Reliable epidemic forecasting is crucial for public health decision-making, especially during pandemics like COVID-19.
• Despite research, accurately forecasting contagion dynamics remains challenging due to complex factors like human behavior and data uncertainty.
• Existing research often struggles to explain the inherent difficulties in predicting disease spread.

Purpose of the Study:

• To rigorously explain the computational difficulty of short-term epidemic forecasting in networked populations.
• To demonstrate the intractability of key forecasting problems using principles from computational complexity theory.
• To identify fundamental limitations in predicting disease parameters and dynamics.

Main Methods:

• Application of computational complexity theory to analyze epidemic forecasting problems.
• Demonstration of intractability for problems like predicting infection numbers and peak times.
• Theoretical analysis, even under ideal conditions with known and stable disease parameters.

Main Results:

• Key epidemic forecasting problems, such as predicting infection counts and peak times, are computationally intractable.
• Solving these problems efficiently would imply a violation of widely held hypotheses in computational complexity.
• Contagion dynamics exhibit fundamental unpredictability at both macroscopic and individual levels.

Conclusions:

• Short-term epidemic forecasting on networked populations faces inherent computational intractability.
• Predicting disease parameters and dynamics presents fundamental challenges, even in simplified scenarios.
• Efficient algorithms and approximation methods have been developed for specific, restricted forecasting problems.