AI-powered drug safety testing is emerging as a transformative approach for early-stage therapeutic evaluation. This method integrates artificial intelligence with experimental models to predict cardiac risk before human trials. The Jackson Laboratory (JAX) has received a contract of up to $30 million from the Advanced Research Projects Agency for Health (ARPA-H) Computational ADME-Tox and Physiology Analysis for Safer Therapeutics (CATALYST) program to develop CARDIOVERSE, an initiative combining stem cell–based systems, genetically diverse models, and computational toxicology. The project aims to generate virtual hearts—digital twins of cardiac tissue—to detect cardiotoxicity earlier, reduce reliance on large-animal testing, and support more confident regulatory decisions.
Developing new drugs is resource-intensive, and cardiotoxicity remains one of the leading causes of clinical trial failure. CARDIOVERSE introduces a comprehensive framework that integrates induced pluripotent stem cell (iPSC) models, mouse systems, high-throughput automation, and AI algorithms to evaluate how diverse patient populations may respond to investigational compounds.
How virtual hearts strengthen computational toxicology
Virtual hearts serve as advanced computational models that simulate electrophysiology, contractility, and metabolic responses. CARDIOVERSE will use stem cell–derived cardiomyocytes and genetically diverse mouse models to capture a wide spectrum of biological variability. These experimental data will train AI systems to predict drug-induced cardiac effects with greater accuracy.
JAX scientists will profile cardiomyocyte cultures using high-throughput automation and imaging platforms. The acquisition of the New York Stem Cell Foundation (NYSCF) enables the team to generate iPSCs at scale through the Global Stem Cell Array, allowing researchers to represent genetic diversity found across patient groups. This integrated workflow creates multimodal datasets that reflect physiologic and molecular changes following compound exposure.
“This is a moonshot vision,” said Matt Mahoney, JAX computational biologist and CARDIOVERSE project lead. “Imagine a future where computational models predict drug safety so reliably that it becomes ethical to move forward on computational evidence alone. That would revolutionize drug development, making it far more affordable and accessible.”
Why cardiotoxicity prediction matters for lab operations
Cardiotoxicity is responsible for late-stage failures and post-approval withdrawals. Traditional approaches rely on animal studies and limited molecular assays that cannot fully capture interindividual differences. Virtual hearts allow scientists to simulate hundreds or thousands of scenarios that represent different genetic backgrounds and potential risk factors.
For laboratory managers, this shift toward AI-powered drug safety testing highlights several operational considerations:
- Increased demand for high-throughput stem cell production
- Integration of automation and robotics into screening pipelines
- Expanded computational capacity for training AI models
- Data governance structures to support large-scale toxicology datasets
- Cross-functional coordination between wet lab teams and informatics groups
As Mahoney noted, computational tools that identify rare but serious reactions before trials begin could reduce costly experimentation and help smaller organizations progress more candidates.
Tools and techniques driving the CARDIOVERSE platform
The initiative incorporates multiple research methods:
- iPSC-derived cardiomyocytes: used to model human cardiac physiology across diverse genotypes
- Mouse models with controlled genetic variation: enabling systematic evaluation of drug response
- High-throughput automation: supporting reproducible cell culture, differentiation, and assay execution
- Molecular and metabolic profiling: capturing gene activity and biochemical shifts following exposure
- AI-based population modeling: creating virtual hearts that simulate differential responses
These tools allow the team to identify biomarkers, mechanistic pathways, and vulnerability patterns tied to drug-induced cardiac stress.
Implications for preclinical strategy
Advanced computational toxicology can improve patient stratification and support regulatory submissions by adding predictive evidence to early-stage studies. By assembling one of the largest cardiotoxicity datasets to date, CARDIOVERSE may help laboratories refine assay selection, update validation standards, and evaluate new AI-enabled screening technologies.
“Too many promising medicines never reach patients because we can’t predict early enough who they will help and who they might harm,” said Lon Cardon, JAX president and CEO. Building virtual hearts that reflect real-world diversity offers a pathway to safer, more targeted therapeutic development.
Lab Management Certificate
The Lab Management certificate is more than training—it’s a professional advantage.
Gain critical skills and IACET-approved CEUs that make a measurable difference.
This article was created with the assistance of Generative AI and has undergone editorial review before publishing.
