Bolek Zapiec | Head of Digital & Data Science, Discovery Development Technologies, R&D
Merck Healthcare
Bolek Zapiec is Head of Digital & Data Science in Discovery Development Technologies, R&D, at Merck Healthcare, where he builds the data and AI architecture connecting discovery science to clinical decision-making. His portfolio spans FAIR data foundations from instrument to warehouse; Synaptix, a knowledge graph serving ~300 scientists across 100+ data sources; and agentic AI systems for evidence synthesis, target assessment, and safety prediction, including HepSAFE, an R&D platform for drug-induced liver injury that serves scientists from the earliest discovery labs through clinical development and regulatory affairs. His architectural instincts are grounded in two decades at the bench. He began in neuroscience at Columbia University, earned his PhD from Heidelberg University, and conducted research at the Max Planck Institute of Biophysics and the Max Planck Research Unit for Neurogenetics, with deep hands-on experience in Bioassay Automation, Multi-OMICS, whole-brain two-photon tomography, STED super-resolution microscopy, and deep learning-based computer vision. At Merck he established Cell Painting as a routine capability, building the infrastructure, pipelines, and phenotypic profiling workflows behind it. He believes the next step change in drug discovery will come less from better models than from better-connected evidence, and builds accordingly.
Appearances:
BioTechX EU Day 3 @ 11:50
HepSAFE: Integrating Clinical, Literature, and Experimental Evidence for DILI Prediction
Drug-induced liver injury (DILI) remains one of the hardest safety problems in drug discovery, for a simple reason: only humans reliably reveal it. Animal and in vitro models are poorly predictive, so the decisive evidence sits in clinical data: fragmented across trials, literature, and internal records, and rarely in a form that supports prospective decisions.
HepSAFE addresses this by treating integration as the primary technical problem. We consolidated heterogeneous clinical, experimental, and literature-derived evidence into a Neo4j knowledge graph, where compounds, findings, mechanisms, and patient-level signals are connected rather than merely co-located. The graph then serves two purposes. First, it provides the structured substrate for training predictive models using primary human hepatocytes and other assays, grounding in vitro readouts in human outcomes. Second, through a conversational interface over the graph, it synthesizes literature, clinical trial data, internal results, and model predictions into answers our drug hunters can actually act on.
HepSAFE is among our first knowledge-graph and AI solutions to serve the full chain: from earliest discovery labs through clinical development and healthcare regulatory affairs. This talk covers the graph design, what worked, and what we would build differently
Get involved at BioTechX Europe
Created by
Terrapinn is proud to be a member of isla.
Working together to build sustainable events