Mantis Biotech's Digital Twins Tackle Medicine's Data Gap

Large language models (LLMs), powered by extensive datasets, hold significant potential to revolutionize biomedical research. Their capabilities span accelerating genomics studies, refining clinical documentation, enhancing real-time diagnostics, bolstering clinical decision-making, expediting drug discovery, and even generating synthetic data to advance experimental designs.

However, the transformative promise of LLMs in this domain often encounters a critical limitation. Beyond the well-structured data commonly found in healthcare, these models frequently struggle with "edge cases," such as rare diseases or unusual conditions, where acquiring reliable and representative data proves exceptionally challenging.

Addressing this crucial data availability gap, New York-based Mantis Biotech asserts it is developing a groundbreaking solution. The company’s platform is engineered to integrate diverse data sources, subsequently generating synthetic datasets. These datasets are then utilized to construct "digital twins" of the human body — sophisticated, physics-based predictive models encompassing anatomy, physiology, and behavior.

Mantis Biotech is positioning these digital twins for wide-ranging applications in data aggregation and analysis. Potential uses include the study and testing of novel medical procedures, training advanced surgical robots, and simulating or predicting medical conditions, and even behavioral patterns. For instance, as Mantis' founder and CEO Georgia Witchel explained to TechCrunch in a recent interview, a sports team could employ these twins to forecast the likelihood of an NFL player developing an Achilles heel injury, factoring in their recent performance, training load, diet, and career longevity.

The construction of these digital twins involves a multi-step process. Mantis' platform initially ingests data from a myriad of sources, including medical textbooks, motion capture cameras, biometric sensors, training logs, and medical imaging. An LLM-based system then routes, validates, and synthesizes these disparate data streams. This consolidated information is subsequently processed through a physics engine, producing high-fidelity renders of the dataset, which are then used to train advanced predictive models.

"We're able to take all these disparate data sources and then turn them into predictive models for how people are going to perform. So anytime you want to predict how a human being is going to be performing, that is a really good use case for our technology," Witchel stated.

Witchel emphasized to TechCrunch that the physics engine layer is pivotal. It significantly enhances the available information by anchoring the generated synthetic data, thereby realistically modeling the complexities of human anatomy and physiology.

"If I asked you to do hand-pose estimation for someone who is missing a finger, it would be really, really hard, because there are no publicly available datasets of labeled hand positions of someone who is missing a finger. We could generate that dataset really, really easily, because we just take our physics model and we say, remove finger X, regenerate model," she elaborated.

Given its capacity to bridge gaps in data sources, Witchel envisions Mantis' platform achieving widespread adoption across the biomedical industry. This is particularly relevant where information on procedures or patients is often difficult to access, unstructured, or siloed across various systems. She specifically highlighted the utility for edge cases and rare diseases, areas where data acquisition is notoriously difficult due to ethical and regulatory restrictions surrounding the inclusion of patient data in public datasets or its use in AI model training.

"You know how when you see a three-year-old running around, and they have a Barbie, and they're holding it by one leg and smashing it against a table? I want people to have that mindset with our digital twins," she remarked. "I think that's going to open up people to this idea that humans can be tested on when you're using virtual humans. I feel currently, people operate with the exact opposite mindset, which totally makes sense, because people's privacy should be respected. In fact, I don't really think people's data should be exploited at all, especially when you have these digital twins."

Currently, Mantis has achieved notable success within professional sports, a sector with a clear demand for modeling high-performing athletes. Witchel confirmed that an NBA team is among the startup's key clients.

"We create these digital representations of the athletes, where it basically shows here's how this athlete has jumped, not just today, but for every single day in the past year, and here's how their jumps are changing over time compared to the amount that they're sleeping, or compared to how many times they lift their arms above their head," she explained, detailing the granular insights provided.

The startup recently secured $7.4 million in seed funding, spearheaded by Decibel VC, with contributions from Y Combinator, several angel investors, and Liquid 2. This capital infusion is earmarked for critical investments in hiring, advertising, marketing, and broader go-to-market initiatives.

Looking ahead, Witchel stated that Mantis' immediate priority is to continue advancing its technology. The ultimate goal is to release the platform to the general public, with an initial focus on preventative healthcare. Furthermore, the company is actively developing solutions tailored for pharmaceutical laboratories and researchers engaged in FDA trials, aiming to provide crucial insights into patient responses to various treatments.