A Clustering Approach for Basket Trials Based on Treatment Response Trajectories

2025-11-13

AI Summary

In basket trials, high heterogeneity in treatment response trajectories across cohorts and reliance on static single endpoints (e.g., ORR) obscure dynamic treatment effects. To address this, we propose a model-free, data-driven clustering framework: similarity between response trajectories is quantified via transition probabilities, and the optimal number of clusters is determined nonparametrically. Subsequently, a hierarchical Bayesian model is fitted to each cluster to enable precise efficacy estimation and inference. Our key innovation lies in abandoning prespecified cluster numbers and static endpoints, instead enabling structural adaptivity through integration of longitudinal response dynamics. Simulation studies demonstrate that the method accurately recovers true cluster structures under high heterogeneity, maintains strict control of Type I error rates, and achieves substantially improved statistical power compared to conventional approaches.

Technology Category

• Machine Learning: Clustering
• Search and Optimization: Sampling/Simulation-based Search
• Cognitive Modeling & Cognitive Systems: Adaptive Behavior

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Metrics for user behavior and evaluating success
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Heterogeneity in efficacy is sometimes observed across baskets in basket trials. In this study, we propose a model-free clustering framework that groups baskets based on transition probabilities derived from the trajectories of treatment response, rather than relying solely on a single efficacy endpoint such as the objective response rate. The number of clusters is not predetermined but is automatically determined in a data-driven manner based on the similarity structure among baskets. After clustering, baskets within the same cluster are analyzed using a hierarchical Bayesian model. This framework aims to improve the estimation precision of efficacy endpoints and enhance statistical power while maintaining the type~I error rate at the nominal level. The performance of the proposed method was evaluated through simulation studies. The results demonstrated that the proposed method can accurately identify cluster structures in heterogeneous settings and, even under such conditions, maintain the type~I error rate at the nominal level while improving statistical power.

Problem

• Groups baskets by treatment response trajectories instead of single endpoints
• Automatically determines cluster count based on basket similarity patterns
• Improves efficacy estimation precision while controlling type I error

Innovation

• Model-free clustering using treatment response trajectories
• Data-driven automatic determination of cluster numbers
• Hierarchical Bayesian model analysis within clusters