Model-robust Inference for Seamless II/III Trials with Covariate Adaptive Randomization

2025-12-10

AI Summary

Existing inference methods for seamless phase II/III trials often rely on restrictive parametric assumptions, are limited to continuous outcomes, and neglect the stratified structure inherent in covariate-adaptive randomization (CAR), thereby limiting practical applicability. Method: We propose a model-robust, unified inferential framework applicable to generalized linear models, accommodating multiple endpoints, hierarchical estimands, and diverse CAR schemes. Our approach explicitly incorporates CAR-induced stratification into variance estimation, integrates Dunnett’s multiple testing procedure with inverse-chi-square combination to strongly control the family-wise Type I error rate, and establishes asymptotic theory via Z-estimation. Contribution/Results: Simulation studies and real-data analyses demonstrate substantial gains in statistical power across binary, count, and continuous endpoints, while maintaining accurate Type I error control. The framework achieves both theoretical rigor and broad practical utility.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Probabilistic Inference
• Intelligent Robots: State Estimation

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Seamless phase II/III trials have become a cornerstone of modern drug development, offering a means to accelerate evaluation while maintaining statistical rigor. However, most existing inference procedures are model-based, designed primarily for continuous outcomes, and often neglect the stratification used in covariate-adaptive randomization (CAR), limiting their practical relevance. In this paper, we propose a unified, model-robust framework for seamless phase II/III trials grounded in generalized linear models (GLMs), enabling valid inference across diverse outcome types, estimands, and CAR schemes. Using Z-estimation, we derive the asymptotic properties of treatment effect estimators and explicitly characterize how their variance depends on the underlying randomization procedure.Based on these results, we develop adjusted Wald tests that, together with Dunnett's multiple-comparison procedure and the inverse chi-square combination method, ensure valid overall Type I error. Extensive simulation studies and a trial example demonstrate that the proposed model-robust tests achieve superior power and reliable inference compared to conventional approaches.

Problem

• Develops model-robust inference for seamless phase II/III trials
• Addresses limitations of existing methods for covariate-adaptive randomization
• Enables valid inference across diverse outcomes and estimands

Innovation

• Model-robust framework using generalized linear models
• Adjusted Wald tests with Dunnett's and inverse chi-square methods
• Valid inference across diverse outcomes and randomization schemes