Score tests based on a finite mixture model of Markov processes under intermittent observation

Area of Science:

• Biostatistics
• Genetics
• Rheumatology

Background:

• Disease progression often follows complex patterns influenced by unobserved factors.
• Intermittent patient examination leads to interval-censored transition times, complicating analysis.
• Identifying genetic associations with disease subtypes is crucial for personalized medicine.

Purpose of the Study:

• To develop a statistical mixture model for analyzing disease progression with latent classes.
• To create a score test for identifying genetic markers associated with these latent classes.
• To apply the model and test to psoriatic arthritis (PsA) data.

Main Methods:

• A mixture model incorporating distinct Markov processes for latent classes.
• Development of a score test for genetic marker association with class membership.
• Simulation studies for algorithm validation and performance assessment.
• Application to PsA data for identifying human leukocyte antigen (HLA) markers.

Main Results:

• The mixture model effectively accommodates varying disease progression pathways.
• The score test successfully identified genetic markers associated with latent disease classes.
• Human leukocyte antigen (HLA) markers were found to be associated with sacroiliac joint damage in PsA patients.

Conclusions:

• The proposed mixture model and score test provide a robust framework for analyzing disease progression and genetic associations.
• The findings highlight specific HLA markers relevant to PsA joint damage, offering potential for targeted therapies.
• The study emphasizes the utility of statistical modeling in understanding complex diseases and their genetic underpinnings.