Sensitivity and Uncertainty Analysis for Two-stream Capture-Recapture Methods in Disease Surveillance

Area of Science:

• Epidemiology
• Biostatistics
• Public Health Surveillance

Background:

• Capture-recapture methods are essential for estimating disease cases in surveillance.
• Existing methods often struggle with data stream dependence and uncertainty quantification.

Purpose of the Study:

• To propose a novel sensitivity and uncertainty analysis framework for capture-recapture methods.
• To enhance the estimation of disease prevalence and incidence using multiple data streams.
• To integrate expert opinion for more robust epidemiological parameter estimation.

Main Methods:

• Multinomial distribution-based maximum likelihood estimation.
• Focus on an epidemiologically interpretable dependence parameter.
• Simulation-based uncertainty analysis incorporating expert opinion.
• Application to HIV surveillance data.

Main Results:

• The proposed framework provides accessible data visualizations for sensitivity analysis.
• It offers a realistic approach to uncertainty analysis by considering expert opinion on nonidentifiable parameters.
• The method facilitates interval estimation and demonstrates reliable performance in simulations.

Conclusions:

• The new framework improves the accuracy and interpretability of capture-recapture estimates in disease surveillance.
• Incorporating expert opinion alongside statistical uncertainty enhances reliability.
• The approach is extendable to more than two data streams, offering broad applicability.