AI-powered programmable virtual humans toward human physiologically-based drug discovery

Artificial intelligence (AI) has generated great interest in drug discovery, but current approaches merely digitize existing experiments, failing to predict clinical outcomes of new compounds. Likewise, pharmacology digital twins, designed for late-phase drug development, lack the ability to bridge translational gaps, limiting their value in early-stage drug discovery. The true potential of AI lies in enabling virtual experiments impossible in the real world, 'testing' novel compounds directly within the human body. Advances in AI, high-throughput assays, and single-cell and spatial omics now enable programmable virtual humans: dynamic, multiscale models that establish a new paradigm of physiologically-based drug discovery. This approach offers a transformative path to evaluate and optimize compound efficacy and safety earlier than ever in the drug discovery process.