Work

From twenty million fragmented data points, to one AI-ready cloud platform

The Opportunity

Lack of clean, structured data is a key factor holding back the AI ambitions of organisations the world over. Siloed libraries, incompatible formats and inconsistent taxonomies produce results that cannot be trusted or relied upon when parsed by AI models, if those models can even ingest the data in the first place. And in regulated industries, where data privacy and safety are of paramount importance, the problems with legacy databases are greatly magnified.
With twenty million ophthalmic images sitting in forty different image storage systems (and 50,000 new images being added every week), Moorfields groundbreaking research was being hampered by availability and access to good data to the extent that producing a useful dataset had become a laborious manual process that could take up to nine months to accomplish.

Our Approach

We built a bespoke, cloud-based data platform that catalogued over 20million images and automatically ingests around 50,000 new scans every week, transforming them into easy-to-use standard formats, which are separated from patient-identifiable data and suitable for interrogation by AI models.
Metadata is pulled in from hospital patient systems, allowing authorised researchers to make specific, targeted requests using natural language prompts.
This solution was designed with information governance at its heart, focused on privacy protection and traceability while still allowing researchers to pull in the necessary metadata to search for images using demographic or diagnostic queries.
Our technical approach to data governance is compliant with stringent anonymity and security requirements, giving Moorfields the needed confidence in data for research study results and patients the confidence that their data is handled securely and their privacy will be protected.

The Impact

The new data platform has enabled researchers to identify study cohorts and receive tailored, de-identified datasets in days, rather than the months the same request might have taken in the past.
This has already led to the publication of breakthrough findings. A group from Deepmind, Moorfields and Google Health used the platform to create an AI model that can predict the progression of retinal disease at least as well as or better than clinicians, opening a clinical window for potential therapeutic treatments.
Pearse A Keane, Consultant Ophthalmologist at Moorfields Eye Hospital NHS Foundation Trust and Professor of Artificial Medical Intelligence at University College London, explains:
“If you looked at the typical types of medical studies before, in ophthalmology, you’d often see a study with 300 patients regarded as a large study. Now, it’s tens of thousands, or hundreds of thousands, or even millions of patients. It wouldn’t have been possible to do a fraction of the things that we’ve been able to achieve without the help of Softwire.”

“ It wouldn’t have been possible to do even a fraction of the things that we’ve been able to achieve without the help of Softwire.

Pearse A Keane, Consultant Ophthalmologist, Moorfields Eye Hospital NHS Trust and Professor of Artificial Medical Intelligence, University College, London.

Our Work

More impactful results