Owkin transformed siloed, unstructured clinical data into a queryable patient registry

First we defined the clinically validated scope with Cleveland Clinic Abu Dhabi (CCAD), focusing on building a patient registry for patient exploration for a single indication - prostate cancer.

Then we deployed our data transformation capabilities to link data from diverse CCAD data sources and harmonize them into a single source of truth. Our AI parsed unstructured documents to automatically structure and extract information.

This allowed Owkin’s team to deploy K Pro on the CCAD secure Azure environment, with CCAD’s preferred reasoning model (GPT-4) - to flexibly query the data, identify patients satisfying inclusion and exclusion criteria and explore their characteristics in seconds.

‍

An AI scientist for clinicians

‍Data Transformation AI integrated CCAD’s electronic health record (EPIC) clinical tables, free-text notes, and 994 pathology PDFs via optical character recognition + LLM extraction into a unified registry of 1,700 prostate cancer patients.

AI-augmented pathology through Owkin's Pathology Explorer Tool analysing 14 digitised H&E prostate cancer slides for cell segmentation and annotation, validated by a CCAD pathologist.

Secure, compliant deployment within CCAD's closed Azure environment using UAE in-region AI models - no data left the hospital, fully compliant with UAE governance rules.

K Pro brought the natural language interface to allow clinicians to query cohort characteristics, treatment patterns, and recurrence outcomes in real time.

‍

K Pro deployment brought tangible results for CCAD