Director Merck West Point, Pennsylvania, United States
At Merck, we have successfully converted 11 years of internal rodent and non-rodent repeat-dose in vivo toxicology data into the SEND (Standard for the Exchange of Nonclinical Data) format. This structured dataset empowers our scientists to explore scientific questions, apply predictive modeling, and identify critical trends in in vivo toxicological findings on a global scale. Here, we present two use cases that highlight the utility of SEND-formatted data in preclinical safety assessments: Unsupervised Relationships: The first use case investigates the associations between in-life measurements—such as aspartate aminotransferase (AST), alanine aminotransferase (ALT), and alkaline phosphatase (ALP)—and histopathological outcomes, including liver necrosis and degeneration. Our analysis of SEND-formatted studies confirmed known relationships, uncovered novel associations, and generated hypotheses regarding the absence of correlations in these areas. Multimodal Data Harmonization: The second use case illustrates the power of harmonizing SEND data with various metadata, including chemical structures, gene expression data, ancillary pharmacology, and histopathology images. This integration enhances our capability to generate chemical and biological hypotheses related to toxicological findings. These hypotheses are actively being investigated for observed toxicology findings in live programs at Merck. Furthermore, the integrated dataset serves as a vital resource for in silico modeling aimed at predicting major organ toxicity labels. Together, these use cases demonstrate the transformative potential of SEND-formatted data in fostering innovation and generating insights throughout the drug research and development process.
