Rhino Health supported a global biopharmaceutical company by curating a cross-jurisdiction multi-hospital optical coherence tomography (OCT) “Federated Dataset” to facilitate drug development.
Image Credit: OpenAI DALL-E
Our client wanted to train an AI model using OCT studies and patient records, including derived longitudinal outcomes. The client also wanted to refine their AI development goals based on knowledge of data available in the ‘real world’. They turned to Rhino Health for help. The client knew Rhino had an extensive global network of hospitals (>30 as of this writing), whose members actively participate in research projects. We facilitated a connection between our client and members of the Rhino Network at leading hospitals across the world to build a high-quality dataset to be used in our client’s research - all without needing to move data. The data diversity afforded by such a global dataset is crucial to developing more generalizable and equitable AI algorithms.
The Rhino FCP includes a “Federated Datasets” application, which makes exploration of and collaboration with data distributed across multiple partners seamless. In this instance, we had OCT images linked to longitudinal Electronic Health Record (EHR) data and clinical notes from hundreds of patients across multiple geographies - all harmonized to the same data model via the Rhino FCP. (The hospital partners will also be able to easily reuse these assets for future research projects, lowering the barriers to collaboration.)
Our client has also been able to take advantage of the Rhino FCP’s Secure Access (aka “Secure Remote Desktop”) Feature. To ensure good performance of their AI model, our client wanted to have consistent annotations for their images. Our client was able to use Secure Access to allow a third party to review & annotate images without ever actually taking possession of the images. The images and the annotations remain behind the custodian’s firewall.
OCT studies provide rich data that can be used for developing sophisticated AI-powered diagnostics and biomarkers, for several therapeutic areas. OCT imaging is non-invasive, unlike other biopsy procedures - making them ideal for evaluating chronic conditions and sensitive tissues. The high-resolution images provide a level of detail that facilitates monitoring even subtle changes in a patient’s condition. OCT features can be quantified as well (e.g. retinal thickness, volume, layer segmentation, and blood flow), allowing for objective monitoring of a patient over time. OCT imaging patterns are also specific to different conditions, allowing for effective diagnosis of conditions such as age-related macular degeneration, diabetic retinopathy, and glaucoma. Combining OCT images with other patient data provides a useful foundation for AI in healthcare.
We are excited to support innovative research using this powerful biomarker, and look forward to facilitating similar future research projects for members of our network.
To learn how to collaborate with members of the Rhino Network for this OCT Federated Dataset, or on other federated research projects, contact us.