Adaptive Subtype and Stage Inference for Alzheimer's Disease

Area of Science:

• Computational biology
• Biomedical data analysis
• Disease progression modeling

Background:

• Event-based models like Subtype and Stage Inference (SuStaIn) capture temporal and phenotypical patterns in progressive disorders.
• Existing models struggle with subtypes exhibiting different progression rates and fixed prior events.
• Understanding disease heterogeneity is crucial for effective treatment strategies.

Purpose of the Study:

• To develop an adaptive algorithm for learning subtype-specific events.
• To enhance subtype and stage inference in progressive diseases.
• To address limitations of current event-based models in capturing variable progression rates.

Main Methods:

• Proposed an adaptive algorithm for joint learning of subtype-specific events and inference.
• Utilized simulation studies to evaluate the algorithm's performance.
• Applied the algorithm to Alzheimer's Disease (AD) data.

Main Results:

• The adaptive algorithm demonstrated improved performance across various metrics in simulations.
• Successfully identified different levels of biomarker abnormality within distinct subtypes of Alzheimer's Disease.
• The method effectively captures disease heterogeneity and progression dynamics.

Conclusions:

• The proposed adaptive algorithm enhances subtype and stage inference for progressive disorders.
• This approach offers a more nuanced understanding of disease heterogeneity and progression.
• Effective for analyzing complex diseases like Alzheimer's Disease with biomarker data.