Cohort-based smoothing methods for age-specific contact rates.

Area of Science:

• Epidemiology
• Mathematical Biology
• Statistical Modeling

Background:

• Social contact rates are fundamental in infectious disease modeling, influencing epidemiological parameters like the reproduction number.
• Accurate quantification of contact patterns is essential for parameterizing dynamic transmission models.
• Existing methods for estimating age-specific contact rates often use piecewise constant or bivariate smoothing, which may not fully capture age-related behavioral changes.

Purpose of the Study:

• To propose and evaluate a novel cohort-based smoothing approach for social contact matrices.
• To incorporate the reciprocal nature of contacts and smooth over diagonals, reflecting age-related behavioral changes.
• To improve the estimation of epidemiological parameters by accounting for cohort effects.

Main Methods:

• Developed a constrained smoothing approach focusing on the diagonal (and subdiagonals) of the social contact matrix.
• Proposed two methods for diagonal smoothing: reordering matrix components and reordering the penalty matrix.
• Employed a likelihood framework with constrained penalized iterative reweighted least squares for parameter estimation.

Main Results:

• A simulation study demonstrated the advantages of the proposed cohort-based smoothing method.
• The methods were successfully applied to Belgian POLYMOD contact survey data from 2006.
• The approach provides a more nuanced understanding of age-specific contact patterns.

Conclusions:

• Cohort-based smoothing offers a more realistic and accurate method for analyzing social contact matrices in infectious disease modeling.
• This technique enhances the reliability of epidemiological parameter estimation and insights into disease transmission dynamics.
• The study provides accessible code for reproducing the results, fostering further research.