The AVERT project is a collaboration between Mersey Care, the UK largest Mental Health Trust, Kings College London (KCL), The University of Liverpool (UoL) and AIMES. The project aims to use AI technologies to predict the onset of crisis and the impact of healthcare interventions for mental health service users.
The project builds upon a very successful feasibility project funded by Innovate UK between 2018 and 2020 and has the potential to transform the treatment and care of mental health patients by predicting and averting crisis rather than by simply reacting. This preliminary work has ingested over 12 years’ worth of NHS clinical data from Mersey Care and generated over 4 billion annotations using natural language processing technology developed by Professor Dobson at KCL. These annotations have been used by Professor Maskill at UoL to create ML algorithms which predict descent into crisis and the interventions most likely to prevent this. The data and analysis has been carried out in a highly secure, NHS accredited facility at AIMES and patients and clinical staff have been directly involved in the project from the outset – in terms of engagement/evaluation around issues such as consent, specificity/sensitivity, usability and trust. The patient and clinician forums have proved fundamental to the progress made to date.
The key tasks to be undertaken during next phase of the project are;
The AVERT project will deliver two important outcomes;
1. A set of clinically evaluated AI algorithms to predict crisis and personalized interventions from unstructured mental health data. These algorithms will be available to the NHS as both web services and as licensed technologies within EPR platform
2. A large-scale training and testing data set, richly annotated which can be used to create predictive algorithms across a range of mental health conditions and treatment interventions.
OpenCARE is a collaboration between AIMES, University College London and Barts Health NHS Trust to better measure heart function using AI in imaging. Working with world leading cardiac imaging and data science expertise, (Prof James Moon and Dr Rhodri Davies) we’re able to demonstrate more precise measurement of the heart, by automatically segmenting MRI scans and providing the results instantly to clinicians and researchers. We have developed AI models around Left Ventricular Ejection Fraction, Arterial Flow Analysis and accessory pathway localisation with more models in the pipeline.
OpenCARE has been trained on over 2m images from 2000 consenting patients across the world.
OpenCARE is available via the OpenCARE Portal or OpenCARE TRE also via our channel partners.
PIBD-SET Quality (Paediatric Inflammatory Bowel Diseases Network for Safety, Efficacy, Treatment and Quality improvement of care ) is a Horizon 2020 funded project, which started in January 2016. The overall goal of this proposal was to develop and validate an algorithm for PIBD, based on high or low risk predictors, which detect early complicated or relapsing disease. This will improve effectiveness, reducing treatment related risks and life-long complications due to uncontrolled disease progression.
AIMES provide a secure and compliant hosting environment, where cohort data is stored and documentation between the European Network can be shared. Project partners include NHS Glasgow, Queen Mary University London, Paris Descartes and Shaare Zedek Medical Center of Isreal.