Applying the Huntington's Disease Integrated Staging System (HD-ISS) to Observational Studies.

Area of Science:

• Neurology
• Biostatistics
• Machine Learning

Background:

• The Huntington's Disease Integrated Staging System (HD-ISS) classifies disease progression into four stages (0-3).
• Implementation challenges include missing data and broad stage definitions.
• Subgrouping within stages may improve precision.

Purpose of the Study:

• To impute HD-ISS stages for observational studies with missing data.
• To define progression subgroups within HD-ISS stages.

Main Methods:

• A machine learning algorithm was employed for stage imputation.
• Agreement between imputed and observed stages was assessed using graphical methods and propensity score matching.
• Progression subgroups were identified using descriptive statistics and optimal cut-point analysis.

Main Results:

• Good overall agreement was found between imputed and observed HD-ISS stages.
• The algorithm showed a tendency to over-assign Stage 0 and under-assign Stage 1 in early progression.
• Imputed stages can be utilized similarly to observed stages in large-scale analyses.

Conclusions:

• Imputed HD-ISS stages are viable for large-scale research, especially when imaging data is unavailable.
• Collapsing early stages (Stage≤1, Stage 2, Stage 3) is a practical alternative when imaging data is absent.
• Defined progression subgroups enhance the identification of homogeneous patient groups.