Search papers, labs, and topics across Lattice.
The paper introduces SHIELD, a method for automated safety signal detection in clinical trials that integrates disproportionality analysis with semantic clustering of adverse event (AE) terms using MedDRA embeddings. SHIELD constructs a utility matrix weighting semantic term similarities by signal magnitude, applies spectral embedding and clustering to identify related AE groups, and annotates clusters with syndrome-level summaries using large language models. The method's ability to recover known safety signals and generate interpretable, cluster-based summaries in a real clinical trial demonstrates its potential to enhance safety assessment and causal interpretation.
SHIELD surfaces interpretable safety signals in clinical trials by fusing statistical disproportionality measures with LLM-annotated semantic clusters of adverse events.
We present SHIELD, a novel methodology for automated and integrated safety signal detection in clinical trials. SHIELD combines disproportionality analysis with semantic clustering of adverse event (AE) terms applied to MedDRA term embeddings. For each AE, the pipeline computes an information-theoretic disproportionality measure (Information Component) with effect size derived via empirical Bayesian shrinkage. A utility matrix is constructed by weighting semantic term-term similarities by signal magnitude, followed by spectral embedding and clustering to identify groups of related AEs. Resulting clusters are annotated with syndrome-level summary labels using large language models, yielding a coherent, data-driven representation of treatment-associated safety profiles in the form of a network graph and hierarchical tree. We implement the SHIELD framework in the context of a single-arm incidence summary, to compare two treatment arms or for the detection of any treatment effect in a multi-arm trial. We illustrate its ability to recover known safety signals and generate interpretable, cluster-based summaries in a real clinical trial example. This work bridges statistical signal detection with modern natural language processing to enhance safety assessment and causal interpretation in clinical trials.
