The AI system NVIDIA® DGX-2^TM has the potential to classify stroke and neurological impairments, determine the underlying causes of cancers, as well as recommend the best treatment for patients.

The NVIDIA® DGX-2^TM AI system is the first phase of the project. In addition to its implementation, King’s will also use the NVIDIA Clara AI toolkit with their own imaging technologies (such as NiftyNet) as well as those from partners including Kheiron Medical, Mirada and Scan Computers. 

This will be the first time in the NHS that federated learning will be applied to algorithm development. It means algorithms can be developed on site using data from each individual hospital, without the need for data to travel outside of its own domain – securing the handling of patient data. 

In an age where security and governance of data is of the highest importance, federated learning is crucial for the development and implementation of AI in clinical environments. AI models will be developed in different NHS Trusts across the UK, built on data from different patient demographics and clinical attributes.

By developing the models at individual NHS trusts, the data will give a more accurate and representative insight into patients from that particular area. The NHS will also be able to combine these trust specific models to build a larger and demographically richer overall model.

Researchers and engineers from King’s and NVIDIA will work together with clinicians on-site at King’s College Hospital, Guy’s and St Thomas’, and South London and Maudsley. The combination of research, technology and clinicians will streamline the discovery of critical data strategies, targeted AI problems and fast-track deployment in clinics.

The development of NVIDIA® DGX-2TM is part of King's ongoing London Medical Imaging and AI Centre for Value-Based Healthcare project which aims to disrupt twelve clinical pathways in oncology, cardiology, and neurology, as well as improve diagnoses and patient care in the NHS.