Intelligent Digital Tools for Screening of Brain Connectivity and Dementia Risk Estimation in People Affected by Mild Cognitive Impairment

Overview

Every three seconds someone in the world develops dementia. There are over 50 million people worldwide living with dementia and by 2030 this figure is expected to reach 82 million. Besides time-consuming patient investigations with low discriminative power for dementia risk, current treatment options focus on late symptom management. By screening brain connectivity and dementia risk estimation in people affected by mild cognitive impairment, the European Union (EU) funded AI-Mind project will open the door to extending the 'dementia-free' period by offering proper diagnosis and early intervention. AI-Mind will develop two artificial intelligence-based digital tools that will identify dysfunctional brain networks and assess dementia risk. Personalised patient reports will be generated, potentially opening new windows for intervention possibilities.

The aim of this study is to validate an AI based risk assessment tool for new clinical neurological data management in five clinical centres (Oslo OUS, Helsinki HUH, Madrid UCM, Rome IRCCS and Rome UCSC)). Today, around 50% of patients with mild cognitive impairment (MCI) are at risk to develop dementia, and that early risk signs include brain network disturbances as an expression of beginning synaptic dysfunction in the course of dementia development. This synaptic dysfunction can be registered by electrophysiological brain signals. The AI-Mind Connector will identify such disturbed brain network based on EEG technology. Brain networks patterns are identified among other mathematical possibilities by Graph theory. Classical machine learning and deep learning approaches of artificial intelligence will be used in automating these brain network identification processes in existing M/EEG data. The secondly developed tool, the AI-Mind Predictor, will serve as an enriched Connector, a multimodal prediction method for risk estimation of dementia in MCI patients. In addition to Connector data, cognitive test results, genetic apolipoprotein E (APOE) allele and P-Tau-protein level information are integrated in the AI-Mind Predictor. The AI-Mind Predictor will discriminate between people at risk for further dementia development and non-at-risk. The anticipated high specific and sensitive AI-Mind Predictor results will be compared to state-of-the-art (SOA) approaches. The cutting-edge AI-Mind model development and testing will be done by available anonymised and prospective pseudo-anonymised data collected at the 5 included clinical centres. Final adaptation, validation, and prototype development will be conducted by the hereby described collection of prospective data of a total 1000 MCI subjects, based on standardized clinical inclusion/exclusion criteria listed below. All patients will sign an informed consent before entering the study. The patients will follow the AI-Mind protocol for a 2-year period in parallel with the SOA follow-up procedures at each hospital and country. The protocol includes repetitive M/EEG measurements, digitalised cognitive testing, and at the first visit a blood sample for APOE allele and p-Tau 181 analyses. At two of our clinical centres (HUH and UCM) clinical MEG is additionally offered for specific feature extraction for modelling by new EEG based AI-Mind Connector technology. Importantly, AI-Mind's new data handling procedure will only use existing well-established, globally accessible and low-cost SOA technologies. With AI-Mind's new data processing approach the goal is to increase today's low predictive value (\<0.5) of SOA clinical dementia prediction, and proactively select, with higher accuracy than before, MCI patients at risk to be able to receive earlier clinical intervention. Thereby, AI-Mind wishes to contribute to delaying dementia development by detecting the risk already at the first visit when symptoms occur.