A new approach for handling longitudinal count data with zero-inflation and overdispersion: poisson geometric process...

Area of Science:

• Biostatistics
• Epidemiology
• Medical Statistics

Background:

• Biomedical time series count data often exhibit simultaneous correlations, clustering, and excess zeros.
• Ignoring these data complexities can lead to inaccurate conclusions in clinical research and treatment outcome assessments.
• Existing models may not adequately address the unique challenges of such data.

Purpose of the Study:

• To develop novel statistical models for analyzing time series count data with complex features.
• To introduce the generalized mixture Poisson geometric process (GMPGP) and zero-altered mixture Poisson geometric process (ZMPGP) models.
• To apply these models to bladder cancer new tumor count data and identify significant covariates.

Main Methods:

• Development of GMPGP and ZMPGP models based on extensions of the geometric process model.
• Implementation of models using Bayesian inference with Markov chain Monte Carlo (MCMC) algorithms.
• Model performance and selection assessed using the Deviance Information Criterion (DIC).

Main Results:

• The developed GMPGP and ZMPGP models effectively handle correlated measurements, clustering, and excessive zeros in count data.
• Application to bladder cancer data demonstrated the models' ability to evaluate trend development and identify influential covariates.
• Bayesian implementation with MCMC provided a robust framework for model fitting and analysis.

Conclusions:

• The GMPGP and ZMPGP models offer improved analytical tools for complex biomedical count time series data.
• Accurate modeling of these data characteristics is crucial for reliable treatment outcome evaluation and covariate identification.
• These advanced statistical approaches enhance the understanding of disease progression and patient-specific factors.