Enabling early prediction and intervention for dementia using AI and routine electronic health record data

​​Dementia risk prediction models have focused on populations at already known high risk, and those in whom non-routine investigations like detailed brain imaging have been undertaken. After multiple failed drug trials, it is increasingly recognised that interventions to reduce dementia risk are required earlier in life before the development of neuropathology. We would like to create a dementia risk prediction model using only routinely collected health data, which could plausibly be deployed at a population level.​