Skip to content
Clinical Trial Events
SCOPE Summit for Clinical Ops Executives

Talk Quality & Monitoring track

Analytics for Automated Outlier Detection of ePRO Data: Ensuring Data Integrity and Quality in Clinical Trials

Tue, Feb 4, 2025 · 5:05–5:25 PM EST Rosen Shingle Creek

About this session

Here we present a method for automated detection of low-integrity ePRO data at the site level. This approach is data-driven and multivariate, acting as a complement to the heuristic and univariate analyses currently done by central monitors in CluePoints. The algorithm converts the ePRO data into numerical format, computes site-level features and detects outlier sites using an ensemble of Empirical-Cumulative-distribution-based Outlier Detection (ECOD) and Density-Based Spatial Clustering of Applications with Noise (DBSCAN).

Speaker

  • John Samuelsson

    Associate Director of AI/ML | Agentic AI Lead · Pfizer

    Speaker