A new cloud data platform to enable ground-breaking healthcare research
Help Moorfields Eye Hospital and its partners take forward their pioneering healthcare research.
Softwire was previously involved in a ground-breaking project with Moorfields Eye Hospital and DeepMind (now part of Google Health), which used machine learning to interpret eye scans and recommend how patients should be referred, as accurately as world-leading doctors. Softwire aggregated and transformed the large dataset required to train the artificial intelligence (AI) algorithm, which included ensuring it was de-identified in line with local and national guidelines.
For this system to be used with patients, it needed to undergo further development and trials. This required additional de-identified datasets, which Softwire was initially asked to produce.
At the same time, the success of the initial project meant a variety of other research teams were looking to start using large, tailored sets of eye scan data to enable or enhance their own studies. One example was AlzEye. Another was a study into the effectiveness of using AI to predict the development of exudative age-related macular degeneration (exAMD). This is one of the more serious forms of AMD, and can result in rapid and permanent sight loss.
We built a cloud-based system to automate the data curation process. The MVP enables us to supply data to researchers more quickly, and provides the groundwork for a future self-service capability that will enable authorised researchers to access tailored, de-identified data on-demand.
The dataset we needed to produce was considerably more complex than the one for the initial study. Broader and deeper data was required, and from more sources. Curating it would be significantly more time-consuming if done manually.
Given the desire for researchers to use this type of large-scale eye scan dataset much more widely, we recommended building foundations that would enable a faster, repeatable process to produce these datasets – and ultimately provide self-service access for authorised researchers.
Laying strategic foundations
To do this, we created a centralised and continually updated store of de-identified eye scans from across Moorfields Eye Hospital’s sites. This contains a variety of metadata required to aid researchers. We built this system in the cloud, in line with NHS guidelines. Delivering it in this way ensures it can provide extremely large datasets in a robust way.
The minimum viable product we created enables those with technical expertise to draw appropriate data from the store, whenever it’s needed for research. The intention is to extend the system to enable self-service by authorised researchers.
Data we curated has been used in pioneering healthcare research projects, while our data platform is enabling researchers to study significantly bigger and more detailed datasets than ever before.
Enabling breakthrough research
The central database of de-identified images and clinical data from across Moorfields Eye Hospital’s sites is now operational. Using this database, we have supplied curated datasets for a variety of research projects, some of which have already published breakthrough findings.
The study to predict the progression of exAMD, for example, was able to create an AI model capable of performing as well as, or better than, clinicians at predicting whether an eye will convert to this condition within the next six months.
Transforming medical research
Beyond the individual research programmes, the ability to automate the creation of these large and complex datasets has had a number of benefits for researchers, which will ultimately benefit patients.
It has dramatically accelerated the speed at which datasets can be produced. This has made certain studies viable, where they would previously not have been. In other studies, researchers can now work with much larger datasets than would otherwise have been possible.
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.
“Softwire’s work to create user interfaces that clinicians and researchers without deep technical expertise can use, will represent another step change in our ability to interrogate the enormous set of data we have.
“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.”
Related Case Studies
Creating new insights that support Britain’s shift to low-carbon electricity
Build the new data insights platform for analysts and decision-makers in the British electricity market, while helping Elexon adopt user-centric, agile delivery approaches.
Helping vulnerable people find safe, long-term housing
Help Change Ahead support many more vulnerable people in their search for safe, long-term housing, by designing and building a technology platform to automate its most resource-intensive tasks.