Turing Biosystems Ltd

Turing Biosystems has developed a highly performant multimodal graph engine natively built to work with biology and AI analytics or digital twin/simulation workloads, able to design and optimise drugs and treatments with a focus on immuno-oncology and rare diseases.

Our product is a highly performant multimodal graph engine natively built to work with biology and AI analytics or simulation workloads. The graph engine is 10 to 100X faster than graph database engines available on the market and can work with low cost infrastructures. It can be used together with LLMs and other AI models or foundational models of biology and clinics.

Our own AI model doesn’t use machine learning but rather a technique called automated reasoning – a type of AI always interpretable by design. It is able to reason rigorously on scientific prior knowledge and real-world evidence data, and can therefore be used adaptably on big or small datasets, heterogeneous datasets, or in precision medicine contexts where there may be only one or a few patients. It is able to predict and simulate all the possible adverse events that can happen in a complex critical system and find ways to avoid those adverse events. Automated reasoning offers audit trails every time, fully revealing and explaining each link in the chain of the reasoning process. Decision-making power remains in the hands of the user, who is the expert on their own patients; the model provides suggestions and explains where those suggestions come from. Its ability to reason on connected multimodal graphs supports multi-modal data integration from diverse sources (genomics, metabolomics, imaging, etc.), making it an ideal tool for clinical translational research and decision-making support.

The platform and engine together are extremely well-suited to use in digital twin/simulation creation. Data processing is the aspect of the digital twin that adds the most significant value. Turing’s AI can do this much faster than previous artificial intelligence models have managed, in large part because of our redesigned FPGA chips, which can be reprogrammed “in the field” to the user’s specifications. Our FPGA designs are proprietary and constitute part of our product. These chips enable extremely rapid and effective computation, making it possible for the Turing AI to complete in hours analyses that might previously have taken days. Additionally, our product works on multimodal data. The aim of a fully integrated bio-digital twin is to have these multiple modes of data represented in one place, just like a living being. Where there is such a gathering, the Turing AI graph engine can reason on all data types simultaneously, creating networks of relationships both between and across data types, improving the understanding of how different elements of a system interact, and enabling predictions about how they might behave or affect each other in the future.

• Immunology/inflammation
• Engineered Biology
• Platform Technology
• Autoimmune diseases
• Oncology
• Metabolic Disorders
• Biomarkers