Large language models (LLMs) hold significant potential for revolutionising genomics research by enhancing clinical documentation, speeding up real-time diagnostics, aiding clinical decision-making, and facilitating drug discovery. They can also create synthetic data to fuel experimental advancements. However, their effectiveness is hampered in specialized areas such as rare diseases, where there is often a scarcity of reliable data.
Mantis Biotech, based in New York, aims to bridge this data gap with its innovative platform that combines various data sources to create synthetic datasets. These datasets are intended to develop “digital twins” of the human body—predictive models that replicate biological systems and behaviours. The application of these digital twins extends to aggregating and analysing data, supporting new medical procedures, training surgical robots, and simulating potential medical issues. For instance, they could predict the risk of an NFL player developing an Achilles injury based on various performance metrics, as explained by Mantis’ founder and CEO, Georgia Witchel.
To create these digital twins, Mantis harnesses data from diverse sources, including textbooks, motion capture, biometric sensors, training logs, and medical imaging. An LLM-driven system processes this data, which is then enhanced through a physics engine to produce high-fidelity models for predictive analytics. Witchel highlighted the platform’s ability to synthesise previously unavailable datasets to aid precise predictions of human performance.
An essential feature of Mantis’ technology is its capacity to generate datasets for unique cases—for example, creating models for individuals with specific physical variations such as missing fingers. This capability can be particularly valuable in biomedical fields where acquiring comprehensive datasets poses ethical and regulatory challenges.
Mantis targets sectors where data is often unstructured or fragmented, particularly in cases involving rare diseases, where patient data accessibility is limited. Witchel noted the potential for digital twins to reshape perceptions in research and testing, advocating for experimentation with virtual models instead of relying heavily on real human subjects to respect privacy concerns.
Currently, Mantis has found traction within professional sports, notably collaborating with an NBA team to model athlete performance. Their platform enables the tracking of individual athletes’ jumping metrics over time, evaluating how variables like sleep and training load affect their performance.
Recently, Mantis secured $7.4 million in seed funding, led by Decibel VC with participation from Y Combinator and other investors. This financial boost will support its hiring, marketing, and operational efforts. Moving forward, Mantis plans to expand its technology and ultimately release it to a broader audience, aiming to enhance preventative healthcare solutions and provide insights for pharmaceutical research during FDA trials.
Fanpage:Â TechArena.au
Watch more about AI – Artificial Intelligence
