SPARCC: Semi-Parametric Robust Estimation in a Right-Censored Covariate Model

Area of Science:

• Statistics
• Biostatistics
• Neurodegenerative disease research

Background:

• Understanding pre-diagnostic symptom changes in Huntington disease is crucial.
• Modeling symptom severity requires handling right-censored 'time of diagnosis' data.
• Existing statistical methods have limitations in efficiency and robustness.

Purpose of the Study:

• To develop a novel, robust, and efficient statistical estimator for analyzing right-censored covariate data.
• To address limitations of current estimators in modeling pre-diagnostic symptom trajectories.
• To introduce the SPARCC estimator and its associated R package for practical application.

Main Methods:

• Proposed the SemiPArametric Robust estimation in a right-Censored Covariate model (SPARCC) estimator.
• Demonstrated doubly robust properties when nuisance parameters are parametrically modeled.
• Showed consistency and semiparametric efficiency with nonparametric or machine learning methods for nuisance parameters.

Main Results:

• The SPARCC estimator demonstrates robustness and efficiency as theoretically predicted.
• Empirical validation using the R package 'sparcc' confirms its claimed statistical properties.
• Successful application to estimate Huntington disease symptom trajectories from Enroll-HD study data.

Conclusions:

• The SPARCC estimator offers a statistically superior approach for analyzing pre-diagnostic Huntington disease symptom data.
• The R package 'sparcc' provides a practical tool for researchers in this field.
• This method enhances the understanding of disease progression in Huntington disease and potentially other neurodegenerative conditions.