Inductive Bio awarded $21M by ARPA-H to improve drug toxicity prediction

26 September 2025

Inductive Bio, an AI drug discovery partner developing virtual chemistry labs, is awarded up to $21M by Advanced Research Projects Agency for Health (ARPA-H) to develop AI-powered human-relevant drug toxicity prediction models. ARPA-H Computational ADME-Tox and Physiology Analysis for Safer Therapeutics  (CATALYST) program aims to revolutionize preclinical drug safety prediction by leveraging human-based models that assess safety of drug candidates more reliably than preclinical animal testing.

Inductive Bio's toxicity‑prediction models will be trained on data derived from advanced human‑based biological systems that capture human-specific physiology - human organoids, ex vivo tissues and organs-on-chip. This mechanistically rich, human-relevant data generated in collaboration with leading pharma and academic institutions, including Amgen, Baylor College of Medicine, Torch Bio, and Cincinnati Children’s Medical Center, is set to greatly improve human-relevance of drug safety assessment, potentially preventing drug-induced adverse events in millions of patients.

Earlier in 2025, Inductive had successfully raised $25M in Series A  financing to further expand its AI tools designed for real-time prediction of absorption, distribution, metabolism, excretion, and toxicity (ADMET) of therapeutic candidates in the human organism. Through gathering of anonymized data from thousands of drug discovery programs in a pre-competitive data consortium, the AI model was trained to capture real-world heterogenous ADMET data and make highly accurate predictions of drug’s ADMET properties. Its Compass and Indy user-friendly software enables to design, run and interpret virtual experiments, addressing the need to make decisions at early stages of the drug development process.

“While animal tests have historically been the gold standard for assessing drug safety pre-clinically, the reality is that they often fail to capture how humans will actually respond to new therapeutic candidates,” said Ben Birnbaum, Inductive’s co-founder and the Principal Investigator for the project team.

Liver plays a crucial role in metabolism and clearance of xenobiotics. Owing in great part to human-specific features of xenobiotics transporters and metabolizing enzymes, bile salt export, composition and metabolism, stress response capacity, and immunity, preclinical animal testing does not accurately predict hepatotoxicity in clinical trials. To date, drug-induced liver injury remains a leading cause of failure in drug development. Additionally, it was estimated that as much as 32% of 222 novel therapeutics approved by the FDA from 2001 through 2010 were affected by serious post-marketing adverse reactions Downing et al., JAMA, May 2017, causing significant morbidity and mortality. In the same manner, inter-species differences in heart electrophysiology and drug metabolism pose a barrier for reliable extrapolation of cardiotoxicity testing results in animals to humans. Drug-induced cardiotoxicity is considered responsible for 16% of total 578 worldwide withdrawals of marketed drugs  Siramshetty et al., Nucl. Ac. Res., Jan 2016.

In the context of CATALYST initiative, Inductive will therefore naturally focus on building AI models from the liver and heart toxicity data and working with the FDA to validate their regulatory applications.