Safety Data Analysis

Safety data have traditionally been summarized as tables and listings that are difficult to review and interpret. In today’s climate of heightened safety concerns, these formats are inadequate for quick and accurate interpretation and comprehension.  Safety Miner provides methods for the rigorous analysis of safety data. Analysis results and reports can be readily deployed for interactive review and for submission / reporting.

• Quantify treatment effects on safety endpoints
• Identify safety signals quickly and accurately
• Select the best dose in early-phase trials
• Identify patients with lab and/or AE issues early in all trials

Safety data pose many challenges to the analyst; there are many endpoints (labs/analytes and adverse events) and comparisons to make, resulting in a testing multiplicity issue. Further, AE count data are sparse, thus complicating the analysis. Safety Miner packages include advanced statistical methods such as Bayesian hierarchical models and inside-out machine learning models that directly address the peculiar safety data issues. These methods provide accurate statistical inference and rapid identification of safety signals.

pRisk plot from hierarchical Bayes analysis of adverse event rates

Improve analysis of safety data and identify safety issues faster

• Rigorous statistical analysis of safety data
• Advanced statistical methods that overcome limitations of traditional methods

Deploy analytic best practices across functional areas

• Reporting of analysis results to safety, epidemiology and clinical study teams
• Deployment of best practices safety analysis throughout organization
• More efficient use of biostatistics resources

Align with regulatory and company mandates

• Analyses supporting periodic and aggregate review of safety data