iCARE: An R package to build, validate and apply absolute risk models

Area of Science:

• Biostatistics
• Epidemiology
• Computational Biology

Background:

• Accurate estimation of absolute disease risk is crucial for personalized medicine and public health.
• Existing tools may lack flexibility in model updating and population tailoring.
• Handling missing covariate data in risk prediction models remains a challenge.

Purpose of the Study:

• To introduce the Individualized Coherent Absolute Risk Estimator (iCARE) R package.
• To provide a flexible and coherent framework for building, validating, and applying absolute risk models.
• To demonstrate the utility of iCARE using a breast cancer risk prediction example.

Main Methods:

• Development of an R package (iCARE) for absolute risk modeling.
• Incorporation of user-defined relative risk models, incidence rates, and risk factor distributions.
• Implementation of a coherent approach for handling missing covariate data via model averaging.
• Inclusion of single nucleotide polymorphisms (SNPs) using odds ratios and allele frequencies.
• Validation of model performance using calibration, discrimination, and risk-stratification metrics on independent datasets.

Main Results:

• The iCARE tool offers flexibility in updating models and tailoring them to specific populations.
• It coherently handles missing risk factor information through model averaging.
• The package supports the integration of genetic data (SNPs) into risk models.
• The validation component allows for rigorous assessment of model performance.

Conclusions:

• iCARE provides a robust and adaptable platform for researchers to develop and apply personalized absolute risk prediction models.
• The tool's features facilitate the rapid incorporation of new knowledge and diverse data types.
• iCARE is a valuable resource for advancing disease risk estimation and personalized risk assessment in various populations.