Early detection of COVID-19 outbreaks using human mobility data

Area of Science:

• Epidemiology
• Public Health
• Data Science

Background:

• Contact-based transmission necessitates mobility restrictions to control COVID-19 spread.
• Effective public health strategies require accurate predictions of outbreak timing, location, and severity.

Purpose of the Study:

• To develop and evaluate predictive models for COVID-19 new cases and test positivity rates.
• To assess the impact of human mobility data on prediction accuracy.
• To categorize COVID-19 outbreak severity for targeted interventions.

Main Methods:

• Analysis of aggregated, anonymized health and cell phone mobility data from Israel.
• Development of predictive models for daily new cases and test positivity rate (7-day forecast).
• Evaluation using Root Mean Squared Error (RMSE) and Magnitude Accuracy (MA).

Main Results:

• Models incorporating mobility data reduced RMSE by 17.3% for new cases and 10.2% for test positivity rate.
• Key predictors included lagged case data, test positivity rates, and internal regional movement.
• Achieved RMSE of 4.812 cases/100,000 and 0.79% for test positivity; MA reached 0.775 and 0.820 respectively.

Conclusions:

• Macro-level human mobility and health data integration enhances COVID-19 outbreak prediction.
• This predictive tool supports data-driven decision-making for targeted public health interventions, especially in the post-vaccination era.
• Aids in mitigating collateral damage from broad restrictions by enabling focused containment strategies.