PROJECTS

PASSIAN

Welcome to the dashboard for the PASSIAN Project - aka Piloting A Secure, Scalable, AI-enhanced Infrastructure for Dementia Research On Real-world Data

Our overarching goal is help combat dementia by building research capability for AI in healthcare using real-world data.

Overview of the project

People are living longer than ever, and so age-related dementias are a key challenge to 21st-century medicine and healthcare. Dementia-causing diseases are complex and poorly understood. Artificial intelligence (AI) models are being developed on artificial research datasets, which are generated in highly controlled, unrealistic scenarios. To truly unlock the potential of AI in dementia research requires access to real-world, diverse biomedical data from the front-line: NHS Memory Clinics. Our goal is to overcome this critical roadblock to AI research by leveraging our clinical connections and data (we run two NHS Memory Clinics). We will use high performance computational resources in combination with our biomedical research expertise (we lead international dementia and AI initiatives). This will allow us to deliver a proof-of-concept, secure, scalable, AI-enhanced infrastructure for dementia research on clinical data. We will also test whether AI confers benefits for enhancing real- world (noisy, low-quality) medical images.

By opening a portal between AI researchers and real-world data from the NHS we will enable and facilitate impactful future AI research in dementia. Piloting a solution to accessing real-world data in a secure research environment is the only way for AI to have a tangible and lasting impact on dementias in the real world.

Main Aims

1	2

pilot a secure, scalable clinical data-sharing solution that will remove a key roadblock to developing AI solutions on real-world biomedical data;	test AI-driven enhancements to clinical imaging data quality. We will start development in one NHS centre, in collaboration with a second, with a view to producing a solution that scales across an entire healthcare system.

Related Projects

QMIN-MC

The aim of the QMIN-MC study is to build a dataset that can be used to validate new artificial intelligence (AI) method in the real world memory clinic setting across the UK. These AI tool uses brain scans and can help in various ways, for example to speed up the diagnosis of memory and thinking disorders, and to help us predict what might happen in the future to people with Alzheimer’s disease and similar diseases. The vision behind this project is that of providing artificial intelligence guided tools making the best use of all the available information from brain scans to support memory clinics in decisions about diagnosis and care.

You can find more info on the official website

Team

Our team features world-leading experience in developing and applying AI and machine learning in dementias. We also work with the Deep Dementia Phenotyping (DEMON) Network (demondementia.com; led by Co-I Llewellyn), which is an international network including over 1,200 academics and clinicians dedicated to collaborative innovation applying data science and AI to dementia research.

Neil Oxtoby, PI	Tim Rittman, PI	David Llewellyn, PI	Zuzana Walker, PI

Neil is based at the Computer Science Department at UCL and is the main lead of the CODEC project	Tim is based at the Department of Clinical Neurosciences at Cambridge University where he leads the QMIN-MC project	David is based at the University of Exeter Medical School and is also the Director of the Deep Dementia Phenotyping (DEMON) Network	Zuzana is based in the Psychiatry Department at UCL and is the lead consultant in the Essex Neurocognitive Clinic (CODEC Project)
Marcella Montagnese, Research Fellow	Bojidar Rangelov, Research Fellow

Marcella is a Neuroinformatics Research Fellow based at the Department of Clinical Neurosciences (Cambridge University) where she works on the development of AI tools for diagnosis and prognosis (QMIN-MC)	Bojidar is a Research Fellow based at the Computer Science Department at UCL where he employs machine learning techniques to analysing CT scan images of patients’ lungs.