Mantis Biotech Creates Human 'Digital Twins' to Tackle Medical Data Shortage

Large language models trained on massive datasets have the potential to accelerate genomics research, streamline clinical documentation, enhance real-time diagnostics, aid clinical decision-making, speed up drug discovery, and even create synthetic data to propel experiments forward.

However, their transformative potential for biomedical research frequently hits a roadblock: beyond the structured data foundational to healthcare, these models falter with edge cases such as rare diseases and atypical conditions, where reliable, representative data is lacking.

New York-based Mantis Biotech asserts it is developing the solution to bridge this data availability gap. The company's platform integrates diverse data sources to produce synthetic datasets, which can then be used to construct "digital twins" of the human body—predictive, physics-based models of anatomy, physiology, and behavior.

The company promotes these digital twins for data aggregation and analysis. They could be employed to study and test new medical procedures, train surgical robots, and simulate or predict medical issues and even behavioral patterns. For instance, a sports team could forecast the probability of a specific NFL player developing an Achilles tendon injury based on recent performance, training load, diet, and career longevity, Mantis founder and CEO Georgia Witchel explained in a recent TechCrunch interview.

To build these twins, the Mantis platform first aggregates data from sources like textbooks, motion capture cameras, biometric sensors, training logs, and medical imaging. It then uses an LLM-based system to route, validate, and synthesize these varied data streams. This information is processed through a physics engine to create high-fidelity renders of the dataset, which subsequently train predictive models.

"We can take all these disparate data sources and transform them into predictive models for human performance. Any scenario where you need to forecast how a person will perform is an excellent use case for our technology," Witchel said.

The physics engine layer is crucial, Witchel told TechCrunch, as it enhances available information by grounding the generated synthetic data and realistically modeling anatomical physics.

"If you needed to estimate the hand pose of someone missing a finger, it would be extremely difficult due to a lack of publicly available, labeled datasets for that condition. We could generate that dataset very easily by simply modifying our physics model—telling it to remove a specific finger and regenerate," she explained.

Because the Mantis platform addresses gaps in data sources, Witchel believes it has broad potential across the biomedical industry, where information on procedures or patients is often difficult to access, unstructured, or siloed. She emphasized its value for edge cases and rare diseases, where data is scarce due to ethical and regulatory constraints on using patient data in public datasets or for AI training.

"You know how a three-year-old might play with a Barbie doll, holding it by one leg and banging it on a table? I want people to adopt that mindset with our digital twins," she said. "I believe it will open people up to the idea that humans can be tested virtually. Currently, the prevailing mindset is the opposite, which is completely understandable—people's privacy must be respected. In fact, I don't believe personal data should be exploited at all, especially when you have these digital twins as an alternative."

Currently, Mantis has found success in professional sports, likely due to the need to model high-performance athletes. Witchel noted that one of the startup's primary clients is an NBA team.

"We create digital representations of the athletes. It shows not just how an athlete jumped today, but how they've jumped every single day over the past year, and how those jumps change over time relative to their sleep patterns or how often they raise their arms overhead," she explained.

The startup recently secured $7.4 million in seed funding led by Decibel VC, with participation from Y Combinator, several angel investors, and Liquid 2. The capital will be used for hiring, advertising, marketing, and go-to-market initiatives.

Witchel stated that Mantis's next steps are to continue developing the technology and eventually release the platform to the public, focusing on preventative healthcare. The company is also working to serve pharmaceutical labs and researchers conducting FDA trials, aiming to provide insights into patient treatment responses.