Inductive Bio, an AI-driven drug discovery company known for developing virtual chemistry laboratories, has been awarded up to $21 million to lead a groundbreaking project aimed at transforming how drug toxicity is predicted. The award, provided under the U.S. Advanced Research Projects Agency for Health (ARPA-H) CATALYST program, will support the development of next-generation AI models designed to improve drug safety assessment while reducing reliance on animal testing.
The initiative, titled DATAMAP (Digital Acceleration of Toxicity Assessment with Mechanistic and AI-driven Predictions), brings together a consortium of leading scientific institutions and industry partners. Inductive Bio will collaborate with Amgen, Cincinnati Children’s Hospital Medical Center, Baylor College of Medicine, and Torch Bio, a University of Michigan Medical School spinout. Together, the team plans to generate biological datasets from cutting-edge human model systems—including organoids, ex-vivo human tissue platforms, and microphysiological systems—to build toxicity prediction tools that more accurately reflect human biology.
The aim is to address a long-standing challenge in drug development: the high failure rate of clinical-stage candidates. Nearly 90% of drugs that enter human trials never reach the market, with approximately 25% failing due to safety issues that were not predicted by traditional preclinical animal models. Inductive Bio and its partners believe that AI trained on advanced human-derived systems could dramatically improve prediction accuracy, mitigating costly failures and enhancing patient safety.
The project will initially focus on two of the most significant causes of drug safety failures: drug-induced liver injury (DILI) and cardiotoxicity. These concerns account for nearly 40% of drug withdrawals due to adverse safety findings. By building high-quality biological datasets and applying sophisticated machine learning methods, Inductive Bio aims to produce AI toxicity prediction models capable of being validated for regulatory decision-making. The company will work with the U.S. Food and Drug Administration (FDA) throughout the process to establish a regulatory framework for the models’ use in investigational new drug (IND) applications.
Amgen will serve as a strategic advisor on how these tools can be integrated into real-world drug development workflows, including their potential role in supporting next-generation therapeutics.
Ben Birnbaum, co-founder of Inductive Bio and Principal Investigator for the DATAMAP project, emphasized the limitations of conventional preclinical testing. “Animal tests have historically been the gold standard, but they often fail to represent how humans will respond to new therapeutic candidates,” he said. “By combining AI with organoids, ex-vivo human tissues, and microphysiological systems, we hope to better predict safety issues that current methods miss. The real-world impact will be demonstrated when our biopharma partner submits an IND using this technology.”
The project also aligns with the FDA’s growing push toward new-approach methodologies (NAMs) as alternatives to animal testing, reflecting a broader industry shift enabled by advancements in human biology modeling and computational science. Support from ARPA-H’s CATALYST program underscores the urgency of modernizing preclinical safety assessments to accelerate the development of safer, more effective medicines.
Once completed, the DATAMAP framework and its AI models are expected to become widely available, enabling drug discovery teams across the biopharmaceutical sector to reduce late-stage failures, lower development costs, and decrease reliance on animal testing—all while helping deliver safer medicines to patients faster.