Sufficient dimension reduction for populations with structured heterogeneity.

Area of Science:

• Statistics
• Health Informatics
• Biostatistics

Background:

• Patient heterogeneity poses challenges for regression models in large populations.
• Comorbidities significantly alter covariate-outcome relationships in health system risk modeling.
• High-dimensional covariates complicate accounting for heterogeneity, even with large datasets.

Purpose of the Study:

• To propose a flexible and interpretable risk modeling approach for handling patient heterogeneity.
• To improve estimation efficiency and interpretability by borrowing strength across subpopulations.
• To develop a robust tool for exploratory analysis and predictive modeling in healthcare.

Main Methods:

• Semiparametric sufficient dimension reduction is employed.
• The approach accounts for patient heterogeneity and related subpopulations.
• Estimation efficiency and interpretability are enhanced through statistical techniques.

Main Results:

• Simulated examples show improved estimation performance with the proposed method.
• The approach demonstrates robustness to violations of underlying assumptions.
• The method effectively models hospital admission risk in a large health system.

Conclusions:

• The proposed semiparametric sufficient dimension reduction approach effectively addresses patient heterogeneity in risk modeling.
• This method offers enhanced accuracy, interpretability, and predictive power for health system analyses.
• The approach serves as a valuable tool for both exploratory and predictive risk modeling.