Artificial intelligence (AI) is rapidly changing how scientists discover and design biomedical materials. In a new review published in Science Bulletin , researchers summarize how AI are accelerating the development of inorganic biomaterials for applications including drug delivery, cancer therapy, anti-inflammatory treatment and tissue engineering.
Inorganic biomaterials, including bioactive ceramics, nanozymes, and metal-organic frameworks, are widely studied for applications such as drug delivery, cancer therapy, anti-inflammatory treatment, and tissue engineering. Many of these materials possess unique optical, catalytic, or structural properties that allow them to regulate reactive oxygen species, deliver therapeutic molecules, or mimic natural enzymes. However, discovering effective materials remains difficult because biological systems are highly complex and material performance depends on many interconnected factors.
“For a long time, biomaterial development has relied heavily on repeated experiments and empirical optimization,” the authors explain. “Researchers often need to synthesize and test large numbers of candidate materials before finding one with suitable biomedical properties.”
According to the review, AI is now being used in two major ways. The first is property prediction, where algorithms analyze existing datasets to predict how a material will behave in biological systems. This includes drug release behavior, toxicity, stability, and interactions with cells and tissues. The second is inverse design, where AI starts from a desired function—such as controlled drug release or anti-tumor activity—and proposes candidate material structures that may achieve that goal.
The review highlights several key applications. In drug delivery systems, AI models are helping predict how different material structures affect drug release rates. In cancer therapy, machine learning is being used to identify nanomaterials with improved catalytic or reactive oxygen species regulation abilities. In inflammatory diseases, AI-assisted screening has enabled the discovery of nanozymes that can reduce oxidative stress and improve therapeutic outcomes. In tissue engineering, AI is being combined with 3D printing techniques to optimize scaffold structures for bone regeneration and tissue repair.
Beyond prediction and design, the study also discusses the emerging role of generative AI models, which can propose entirely new material structures that have not been previously observed. Large-scale computational tools, such as graph neural networks and foundation models, are further expanding the searchable chemical space, allowing faster discovery of promising candidates.
Despite the rapid progress, the review also points to several major challenges. One major issue is data quality and consistency, as experimental datasets often vary across laboratories. Another challenge is model interpretability, since many AI systems function as “black boxes” that do not clearly explain their predictions. In addition, translating computational results into real biomedical applications still requires extensive experimental validation.
Even so, the review suggests that AI could significantly reshape the future of biomaterial discovery. By integrating machine learning, automated experiments, and high-throughput screening, future research platforms may become faster, more predictive, and increasingly autonomous.
According to the authors, the long-term vision is to establish an AI-driven materials discovery pipeline that can connect clinical needs with material design, ultimately supporting the development of safer and more effective therapies.
Science Bulletin
Systematic review