Carl Cohen once said, “The use of animals in research is not a choice between humans and animals, but a choice for human health and safety.” Animal studies in biomedical research provide essential data for safety and efficiency. This allows for the development of vaccines, therapies, and interventions. By using animals such as mice, rats, and non-human primates, researchers can examine complicated biological processes, study disease development, and assess possible treatments, which would be impossible or unethical in humans (Mukherjee, 2022). These models allow for controlled testing conditions, involving diet, genetics, and exposure to bacteria and viruses, which improves accuracy and replication. The Food and Drug Administration (FDA) relies largely on preclinical animal data to determine the safety and effectiveness of new drugs before proceeding to human clinical trials. Furthermore, animal studies permit scientists to look into long-term results and systemic effects, offering essential knowledge into persistent diseases and therapeutic effects. Overall, animal research acts as a connection between laboratory discoveries and clinical applications.

Animal studies have multiple key advantages that make them essential in biomedical research. Biological similarities to humans allow scientists to simulate complicated diseases, test feasible therapies, and determine treatment outcomes with suitable accuracy (Thompson-Iritani, 2025). By using whole-body system examinations, researchers can see how different organs and physiological systems interact instead of relying only on in vitro models or cell cultures. Effective screening in animals is essential for identifying toxic reactions, identifying proper dosing, and reducing risks before human trials (Hannes, 2024). The FDA uses these animal-derived data to determine regulatory choices and protect public health. In addition, animal studies allow for experiments that would be unethical in humans, such as testing high-dose toxicity or inducing specific diseases. These advantages combined demonstrate why animal research remains a fundamental component of biomedical research.

In spite of these strengths, animal models have major limitations that researchers and enforcement agencies like the FDA must take into account. (Thompson-Iritani, 2025). Poor human comprehension is a major concern, as species-specific differences in metabolism, physiology, and disease development can result in insufficient results. As ethical issues remain with animal experimentation, it has caused public distrust due to animal suffering (Hannes, 2024). Maintaining animal models is extremely expensive and resource-intensive, calling for specialized housing, veterinary care, and strict regulatory enforcement. Some human conditions, such as psychiatric or complex diseases, would be challenging to accurately duplicate in animals, which limits their predictive worth for clinical outcomes (Mukherjee, 2022). The FDA recognizes that while animal studies are necessary for early safety and efficacy testing, they cannot replace human clinical trials. Understanding these limitations has made it possible to apply animal research responsibly, striking a balance between the benefits of science and ethical and translational issues.

The advancement of toxicology is based on the ability of scientists to copy human biology in laboratory systems rather than relying on animal testing. Traditional animal systems usually fail to predict human-specific responses due to interspecies differences in metabolism, gene expression, and physiology. New Approach Methodologies (NAMs) address these issues by combining engineered biological systems with computational tools to advance human significance. Kwon (2026) identifies a shift toward platforms such as induced pluripotent stem cell (iPSC)-derived organoids, microfluidic organs-on-chips, and artificial intelligence (AI)-based models. These methods raise predictive accuracy and are consistent with toxicological concepts, including dose-response relationships and human-specific absorption, distribution, metabolism, and excretion (ADME).

Microfluidic organs-on-chips and three-dimensional organoids utilize human induced pluripotent stem cells (iPSCs) to generate physiologically relevant models by combining human genetic identity with controlled cellular environments. iPSCs keep donor-specific genomes and can differentiate into specialized cell types through signaling pathways that direct lineage commitment, resulting in tissues such as liver, heart, and neural structures. In organoids, cells self-organize into three-dimensional architectures that replicate tissue organization and cell–cell interactions. In contrast, organs-on-chips employ microfluidic flow to create dynamic environments with shear stress, nutrient gradients, and mechanical cues that regulate gene expression and function. These characteristics enable organ-specific mimicry, such as hepatocyte metabolism in liver models or electrical conduction in cardiac tissues. For example, fialuridine (FIAU) induced liver toxicity in humans but not in animal models; human iPSC-derived liver systems successfully reproduced this effect. Collectively, these platforms offer high-fidelity human simulation, supporting more accurate drug testing, accelerated screening, and patient-specific modeling.

Computational models and generative AI systems now predict toxicological outcomes such as skin sensitization and liver injury using chemical databases, in vitro assay data, molecular descriptors, and human-relevant datasets instead of live-animal testing. Quantitative Structure–Activity Relationship (QSAR) models relate chemical structure to biological activity by identifying molecular features that correlate with toxicity endpoints, such as electrophilic reactivity linked to skin sensitization. Deep learning and multi-task neural networks capture complex, non-linear relationships across integrated datasets to predict multiple toxic effects at once. Generative AI expands on these approaches by learning patterns from large datasets to simulate molecular interactions and estimate the safety of novel compounds before synthesis. In practice, regulatory agencies and pharmaceutical companies apply these tools to screen compounds for drug-induced liver injury (DILI), using metabolic and clinical data to anticipate hepatotoxicity, while skin sensitization models predict immune responses based on protein binding potential. Interpretable AI methods highlight the features driving predictions, improving transparency, but performance depends on data quality, coverage, and validation (Zhu, 2020). 

Reductionist New Approach Methodologies (NAMs), including organoids and organs-on-chips, face biological and technical limitations that prevent full replication of complex systemic processes. These systems lack systemic context and multi-organ axis crosstalk, such as interactions between the liver, gut, and endocrine organs that regulate metabolism and homeostasis. Many models omit key cell types, structural architecture, and supporting stromal or immune components, limiting accurate immune system interactions and preventing features like immunosenescence and tissue aging. Inadequate vascularization restricts nutrient and oxygen delivery, reducing long-term viability and impairing metabolic competence (ADME), including absorption, distribution, metabolism, and excretion processes. Materials such as polydimethylsiloxane (PDMS) can absorb hydrophobic compounds, altering drug concentration and experimental outcomes. These systems also struggle with heterogeneity and scalability, making it difficult to reproduce consistent results across platforms. As a result, endocrine signaling, whole-organ interactions, and age-related physiological changes remain incompletely modeled (Zhang, 2021).

In conclusion, animal studies keep being a vital tool in biomedical research, closing the gap between laboratory discoveries and clinical applications. They provide controlled, high-level, whole-organism data for safety and efficacy testing, enabling accelerated research, disease modeling, and preclinical evaluation of drugs and therapies. However, difficulties such as species distinctions, ethical issues, high costs, and limited human comprehension require careful observation. Agencies like the FDA depend on animal studies while stressing the need for complementary human trials to confirm findings. By recognizing the strengths and limitations of animal models, researchers can responsibly use these studies to advance medical knowledge, improve patient safety, and foster innovation. With ethical oversight and strict experimental methodology, animal research remains vital for establishing safe and useful biomedical treatments. “Without animal research, many life-saving treatments and vaccines would never have been developed; it remains an essential step in translating science into medicine.” – National Institutes of Health. New approach methodologies represent a significant shift toward human-centric and predictive toxicology. Induced pluripotent stem cell (iPSC)-derived organoids and organs-on-chips replicate essential structural and functional characteristics of human tissues. Computational models and artificial intelligence facilitate rapid, data-driven toxicity predictions. Empirical evidence demonstrates that NAMs can identify risks that animal models may overlook. For instance, the OECD QSAR Toolbox employs quantitative structure-activity relationship (QSAR)-based tools for predicting skin sensitization, while liver-on-chip systems have revealed the toxicity of fialuridine. Nevertheless, NAMs face limitations, particularly in modeling systemic interactions, immune responses, and aging, which currently prevent them from serving as a complete alternative to animal testing. According to Kwon (2026), the future of toxicology will rely on integrating experimental and computational NAMs to generate reliable and human-relevant data. Ongoing advancements in data quality, integration, and validation frameworks will be essential to enhance predictive accuracy and promote broader regulatory acceptance of NAMs.

References

Fanizza, C., Campanile, M., Forloni, G., Giordano, R., & Albani, D. (2022). Induced pluripotent stem cells in disease modeling and drug discovery.

Hannes Kahrass, Pietschmann,I., & Mertz, M. (2024). Why Do I Choose an Animal Model or an Alternative Method in Basic and Preclinical Biomedical Research? A Spectrum of Ethically Relevant Reasons and Their Evaluation. Animals, 14(4), 651–651https://www.mdpi.com/2076-2615/14/4/651

Kleinstreuer, N. C., & Hartung, T. (2024). Advances in computational toxicology and new approach methodologies.

Kwon, D. (2026). The age of animal experiments is waning. Where will science go next? Nature. https://www.nature.com/articles/d41586-026-00563-3

Mukherjee, P., Roy, S., Ghosh, D., & Nandi, S. K. (2022). Role of animal models in biomedical research: A review. Laboratory Animal Research, 38(1). https://link.springer.com/article/10.1186/s42826-022-00128-1

Sung, J. H., Wang, Y. I., Narasimhan Sriram, N., Jackson, M., Long, C. J., Hickman, J. J., & Shuler, M. L. (2018). Microfabricated human organs-on-chips for disease modeling and drug development.

Thompson-Iritani, S. A., & Newsome, J. T. (2025). Animal research at a crossroads: Strengths, weaknesses, opportunities, and emerging threats. Journal of the American Association for Laboratory Animal Science, 1–5. https://pmc.ncbi.nlm.nih.gov/articles/PMC12379634/