Assessing reliable human mobility patterns from higher order memory in mobile communications.

Area of Science:

• Computational social science
• Epidemiology
• Network science

Background:

• Human mobility patterns are crucial for applications like epidemic forecasting and migration studies.
• Mobile phone records are widely used to proxy human mobility, revealing memory in movement.
• The exact role of memory in these mobility proxies remains unclear.

Purpose of the Study:

• To evaluate the accuracy of standard Markovian mobility models using mobile phone data.
• To introduce and validate an adaptive memory-driven approach for human mobility modeling.
• To assess the impact of mobility modeling accuracy on epidemic spread predictions.

Main Methods:

• Analysis of 560 million call detail records from Senegal.
• Comparison of standard Markovian models (including higher-order) with an adaptive memory-driven model.
• Evaluation of model performance in capturing conditional waiting times and diffusion rates.

Main Results:

• Standard Markovian models fail to capture real mobility patterns and introduce spurious movements.
• The adaptive memory-driven approach realistically models conditional waiting times based on historical movements.
• Individuals in standard models diffuse faster than observed, while the adaptive model shows strong agreement with real data.

Conclusions:

• Accurate modeling of human mobility, incorporating memory, is essential for realistic predictions.
• Standard mobility models can lead to inadequate estimations of disease incidence and geographical spread.
• The adaptive memory-driven approach offers a more reliable method for understanding and predicting human movement, with implications for public health policy.