AI SCIENCE
AI can now build a fake version of your body
LLMs could help speed up genomics research, improve clinical paperwork, support diagnoses, and even help with drug discovery.
But there’s a catch. In healthcare, these models work best when lots of structured data is available.
They struggle more with rare diseases and unusual conditions, where good data is limited.
Mantis Biotech, a startup based in New York, says it is building a way around that.
Its platform combines different data sources to create synthetic datasets that can be used to build “digital twins” of the human body.
These are physics-based models designed to simulate how the body works.
The company says these digital twins could be used to test procedures, train surgical robots, and predict medical issues or physical performance.
For example, they could be used to estimate an athlete’s risk of injury based on things like training load, sleep, diet, and past performance.
To build these models, Mantis uses data from sources like medical imaging, biometric sensors, motion capture, training logs, and textbooks.
An LLM-based system helps sort and combine the information, while a physics engine makes the final model more realistic.
What to know:
Mantis Biotech is building digital twins of the human body using synthetic data.
The goal is to fill gaps in healthcare data, especially in rare or unusual cases.
The company is starting in sport but wants to expand into healthcare and drug research.
The twin thing
That physics layer is especially useful in cases where real-world data is hard to find.
Mantis says it can generate synthetic data for rare cases that are not well represented in public datasets.
The company believes this could be useful across healthcare, especially where data is hard to access, unstructured, or limited by privacy and regulation.
So far, Mantis has seen early success in professional sport.
Founder and CEO Georgia Witchel said one of its main clients is an NBA team using digital models to track athlete performance over time.
Mantis recently raised $7.4 million in seed funding led by Decibel VC, with backing from Y Combinator, Liquid 2, and angel investors.
The company plans to keep building the platform and expand into preventative healthcare, pharma research, and clinical trials.
Sign me up - you have my permission to use my digital twin as your guinea pig. - MV