The design of a panel study under an alternating Poisson process assumption

Area of Science:

• Statistics
• Biostatistics
• Epidemiology

Background:

• Estimating average durations in dynamic processes is crucial for understanding system behavior.
• Alternating Poisson processes model systems with two alternating states, common in health and reliability studies.
• Panel studies are valuable for longitudinal data collection but require efficient design.

Purpose of the Study:

• To develop and evaluate optimal panel study designs for estimating average state durations in alternating Poisson processes.
• To compare different panel study designs based on follow-up wave numbers and total follow-up length.
• To provide practical guidance for applying these designs to real-world data, such as morbidity data.

Main Methods:

• Examined two panel study designs: fixed number of follow-up waves with variable total length, and variable number of waves with fixed length.
• Derived simple expressions for nearly optimal designs.
• Compared proposed designs with continuous observation scenarios.
• Analyzed both equilibrium and nonequilibrium cases.

Main Results:

• Presented simple expressions for nearly optimal panel study designs.
• Demonstrated that specific designs can efficiently estimate average state durations.
• Showcased the applicability of the methods through a morbidity data example.

Conclusions:

• The proposed panel study designs offer efficient methods for estimating average durations in alternating Poisson processes.
• The findings provide a framework for optimizing longitudinal study designs in various fields.
• The application to morbidity data highlights the practical utility of these statistical approaches.