Mantis Biotech is making ‘digital twins’ of humans to help solve medicine’s data availability problem

Mantis Biotech is developing a platform that creates “digital twins” of humans—physics-based, predictive models of anatomy, physiology, and behavior—by integrating and synthesizing data from diverse sources such as textbooks, motion capture, biometric sensors, training logs, and medical imaging. Using a large language model (LLM)-based system combined with a physics engine, the platform generates high-fidelity synthetic datasets that fill gaps in real-world data, especially in rare or edge cases where reliable data is scarce or ethically constrained. These digital twins can be used for various biomedical applications, including studying medical procedures, training surgical robots, simulating medical conditions, and predicting outcomes such as injury risk in athletes. The key innovation lies in the physics engine, which grounds the synthetic data in realistic anatomical and physiological models, enabling the creation of datasets that do not exist in public repositories—for example, modeling hand poses for individuals missing fingers. Mantis aims to address the biomedical industry’s challenge of fragmented, siloed, or