New York-based startup Mantis Biotech is tackling the challenge of limited medical data availability by creating ‘digital twins’ of the human body through synthetic datasets. These digital twins are predictive models of anatomy, physiology, and behavior, generated by integrating diverse data sources like textbooks, motion capture cameras, and medical imaging.
Mantis Biotech’s approach aims to address the limitations faced by large language models in handling edge cases, such as rare diseases where reliable data is scarce. The company’s platform leverages these digital twins for applications like data aggregation, analysis, testing new medical procedures, training surgical robots, and simulating medical scenarios. For instance, the technology could help predict the likelihood of sports injuries based on an athlete’s performance data.
By utilizing an LLM-based system to synthesize and validate data streams, Mantis Biotech’s platform enables the creation of high-fidelity predictive models. This innovation has the potential to revolutionize genomics research, clinical decision-making, drug discovery, and medical experimentation.
Source: TechCrunch