Knowledge-based Graphical Method for Safety Signal Detection in Clinical Trials
Francois Vandenhende, Anna Georgiou, Michalis Georgiou, Theodoros Psaras, Ellie Karekla, Elena Hadjicosta
TL;DR
This work tackles the bottleneck in safety review where MedDRA's hierarchy obscures semantic and mechanistic relationships among adverse events. To address this, it introduces a hidden medical knowledge layer, Safeterm, that encodes MedDRA PTs into high-dimensional embeddings and a decoding toolbox for semantic clustering, a 2-D interpretability map, and indication-specific expectedness scoring. Disproportionality is assessed with EBGM at both PT and cluster levels, with $E_i = N_i \cdot \frac{\sum_i n_i}{\sum_i N_i}$ and $EBGM_i = \frac{n_i + \alpha}{E_i + \beta}$, and cluster EBGM computed as $\mathrm{EBGM}_{cluster} = \frac{\sum_j w_j \mathrm{EBGM}_j}{\sum_j w_j}$ where $w_j = 1/\mathrm{Var}_{EBGM,j}$ and $\mathrm{Var}_{EBGM,j} = \mathrm{EBGM}_j^2/(n_j + \alpha)$. Applied to three legacy trials, the method recovered expected safety signals and produced intuitive visuals (Safeterm map and EVD plot) that enhance interpretability and efficiency. An online Safeterm app linked to ClinicalTrials.gov supports interactive safety review in practice.
Abstract
We present a graphical, knowledge-based method for reviewing treatment-emergent adverse events (AEs) in clinical trials. The approach enhances MedDRA by adding a hidden medical knowledge layer (Safeterm) that captures semantic relationships between terms in a 2-D map. Using this layer, AE Preferred Terms can be regrouped automatically into similarity clusters, and their association to the trial disease may be quantified. The Safeterm map is available online and connected to aggregated AE incidence tables from ClinicalTrials.gov. For signal detection, we compute treatment-specific disproportionality metrics using shrinkage incidence ratios. Cluster-level EBGM values are then derived through precision-weighted aggregation. Two visual outputs support interpretation: a semantic map showing AE incidence and an expectedness-versus-disproportionality plot for rapid signal detection. Applied to three legacy trials, the automated method clearly recovers all expected safety signals. Overall, augmenting MedDRA with a medical knowledge layer improves clarity, efficiency, and accuracy in AE interpretation for clinical trials.