SECRET: Semi-supervised Clinical Trial Document Similarity Search

2025-05-16

AI Summary

Clinical trial design faces challenges in efficiently retrieving semantically similar historical trials to mitigate safety and operational risks—such as severe adverse events or patient recruitment difficulties. Method: We propose the first semi-supervised similarity retrieval framework tailored for clinical trial protocols. It integrates domain-adaptive pretraining, protocol structure-aware semantic summarization, contrastive learning–driven dual-encoder retrieval architecture, and semi-supervised pseudo-label optimization. The framework supports protocol-level partial matching and zero-shot patient-to-trial matching. Contribution/Results: Compared to the strongest baseline, our method achieves 78% higher recall@1 and 53% higher precision@1. It significantly outperforms existing approaches on partial matching and zero-shot tasks, overcoming key limitations of conventional full-text search and supervised fine-tuning paradigms.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Semi-Supervised Learning
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Clinical trials are vital for evaluation of safety and efficacy of new treatments. However, clinical trials are resource-intensive, time-consuming and expensive to conduct, where errors in trial design, reduced efficacy, and safety events can result in significant delays, financial losses, and damage to reputation. These risks underline the importance of informed and strategic decisions in trial design to mitigate these risks and improve the chances of a successful trial. Identifying similar historical trials is critical as these trials can provide an important reference for potential pitfalls and challenges including serious adverse events, dosage inaccuracies, recruitment difficulties, patient adherence issues, etc. Addressing these challenges in trial design can lead to development of more effective study protocols with optimized patient safety and trial efficiency. In this paper, we present a novel method to identify similar historical trials by summarizing clinical trial protocols and searching for similar trials based on a query trial's protocol. Our approach significantly outperforms all baselines, achieving up to a 78% improvement in recall@1 and a 53% improvement in precision@1 over the best baseline. We also show that our method outperforms all other baselines in partial trial similarity search and zero-shot patient-trial matching, highlighting its superior utility in these tasks.

Problem

• Identifying similar historical clinical trials for better trial design
• Improving trial efficiency and patient safety through protocol analysis
• Enhancing similarity search accuracy for clinical trial documents

Innovation

• Semi-supervised learning for clinical trial similarity
• Summarizing protocols to enhance similarity search
• Superior recall and precision over baselines