A New York-based startup, Mantis Biotech, is working on a breakthrough technology that could significantly reshape how healthcare research and sports science operate. The company is tackling a long-standing problem in biomedical research — the lack of high-quality data for rare diseases and unusual medical conditions.
While artificial intelligence, especially large language models, has already improved areas such as clinical documentation and diagnostics, these systems often struggle when dealing with limited or missing datasets. This is particularly true for rare diseases or edge cases, where reliable data is either scarce or unavailable. Mantis Biotech aims to solve this issue by creating “digital twins” of the human body — highly detailed, physics-based virtual models that simulate human anatomy, physiology, and behavior.
The platform developed by Mantis aggregates data from multiple sources, including medical textbooks, imaging systems, biometric sensors, motion capture devices, and training logs. It then uses an AI-driven system to organize and validate this data before passing it through a physics engine. This process generates high-quality synthetic datasets that can be used to simulate real-world human conditions with impressive accuracy.
One of the key advantages of this approach is the ability to generate data that does not exist in reality. For example, if researchers need to study hand movements of individuals missing a finger — a scenario with almost no publicly available datasets — the system can simulate such conditions using its physics-based models. This makes it possible to fill critical gaps in medical data and train AI systems more effectively.
According to Mantis founder and CEO Georgia Witchel, the technology is particularly useful for predicting how the human body will perform under different conditions. One of the company’s early applications is in professional sports, where it is already working with an NBA team. By creating digital replicas of athletes, the system can analyze metrics like jump performance, sleep patterns, and training intensity to identify injury risks.
For instance, the platform can predict the likelihood of an Achilles injury by analyzing how an athlete’s performance changes over time in relation to their physical load and recovery patterns. This kind of predictive insight could help teams prevent injuries before they happen, potentially saving careers and reducing costs.
Beyond sports, the implications for healthcare and pharmaceuticals are substantial. Digital twins could enable safer testing of medical procedures, assist in training surgical robots, and accelerate drug discovery. They also provide a way to study rare diseases without relying on sensitive patient data, which is often restricted due to ethical and regulatory concerns.
Witchel describes the concept as allowing researchers to experiment freely in a virtual environment, similar to how a child tests ideas while playing. By shifting experimentation from real patients to simulated models, the technology offers a safer and more ethical alternative for medical innovation.
To support its growth, Mantis Biotech recently secured $7.4 million in seed funding. The round was led by Decibel VC, with additional backing from Y Combinator, Liquid 2, and several angel investors. The company plans to use this funding to expand its team, strengthen its marketing efforts, and scale its platform.
Looking ahead, Mantis aims to make its technology accessible to a wider audience, with a particular focus on preventative healthcare. It is also developing tools tailored for pharmaceutical companies to assist in FDA trials, offering insights into how patients might respond to treatments through simulation.
If successful, Mantis Biotech’s digital twin technology could mark a major shift in how medical research is conducted — enabling faster, safer, and more data-rich innovation across the healthcare industry.