Led by Dr Alejo J Nevado-Holgado, Oxford University, in association with the University of Manchester and DeepCognito Ltd
The NewMind funded Stage 1 Feasibility Study led by Kim et al (Unlocking the evidence from electronic patient records for smart intervention of mental health disorders; a case study in Alzheimer’s Disease) demonstrated the concept was possible in one CRIS centre (Oxford), extracting dementia diagnosis, 3 medications and 2 mental health scores, obtaining F1 accuracies ranging from 81.48% to 98.05%. These variables were then used to epidemiologically study subtypes of dementia and drug effects. To demonstrate scalability, this Stage 2 study aims at expanding capability to a second UK-CRIS centre (London Slam), a second prominent disorder (depression, besides Alzheimer disease), a larger dictionary of variables (6 drugs, 4 scores, 3 symptoms), and 2 socio-economic variables (e.g. EQ-5D) used by NHS to decide medical practise. These variables will then be used in translatable epidemiological analyses and to estimate best-personalised treatment per individual patient with neural networks.