Our Oncology Diagnostics Inference Network - ODIN™

Using our ODIN platform we are creating the world’s largest purpose-built 3-D cell culture data set to predict the best drug for every cancer patient and create the best drug for every cancer. 


From patient tumor samples, we generate hundreds of micro-tumors and treat them with an extensive panel of drugs or drug combinations.


Through machine-learning-based image analysis, we determine drug sensitivity in 5-7 days. Our data set grows by thousands of images with each patient sample allowing us to derive increasingly sophisticated and precise biological, computational and clinical insights. Data can be provided to oncologists for use as a decision support tool or in partnership with drug developers for biomarker profiling, clinical trial optimization or companion diagnostics development.

The platform comprises fully integrated and automated wet lab and data science capabilities.

M3DUSA™ Models - Micro 3-D Universal Support Architecture

Our biomaterials and cancer biology teams have developed and optimized proprietary 3D cell culture models to faithfully recreate the tumor microenvironment - M3DUSA™ models. We take patient cancer samples and use them to create hundreds of M3DUSA™ models in our automated, high-throughput laboratory. We give these micro-tumors everything they need to behave the way they would in a patient's body, including using immune and scaffolding cells from the cancer patient. 


Each model is treated with a different drug or drug combination and at multiple dose levels to determine the best treatment for that patient-specific tumor.

Our data science and software engineering teams have built a state-of-the-art analysis pipeline to collect and process the vast amounts of data the M3DUSA™ models provide. Our high-dimensional imaging outputs are incredibly data rich and we use machine learning to look for phenotypic and biochemical indicators of real tumor sensitivity, many of which are not discernible through human observation or other traditional methods.


This true collaboration of life science and data science provides unique and unparalleled insights to clinicians and researchers alike.

IRIS™ - Image Representation Insight System

The ODIN™ platform provides patient-specific treatment recommendations in less than a week.


In contrast, current approaches that use genetic markers as surrogate predictors of possible tumor response are not truly personalized predictions but are rather based on large population studies and pooled historical data - the antithesis of personalization.

The ODIN™ platform provides a functional readout of how a patient’s actual tumor responds to treatments in a clinically relevant timeframe.


Current genetics-only based approaches provide only anticipated response or pure prediction, which studies have shown to translate into clinical benefit in only 5% of cases. 

The ODIN™ platform can be used for any solid tumor.


In contrast, genetics based approaches are only available for a very small subset of patients for whom an actionable mutation has been identified, and even for those few patients, such approaches offer little benefit.