First Alzheimer’s Association Academy of 2024 addresses the topic of AI and dementia On 23 January, we hosted our first Alzheimer’s Association Academy meeting of the year. These capacity-building workshops bring together representatives of national Alzheimer’s associations with members of the European Working Group of People with Dementia and European Dementia Carers Working Group, to learn about the latest advances in dementia research, policy, care and treatment from experts in those fields. The Academy meeting, which was moderated by Angela Bradshaw (Director of Research and Policy), was focused on the topic of Artificial Intelligence (AI) and dementia, and welcomed over 70 registrants from 18 countries.
The first two speakers, Holger Frohlich (Fraunhofer SCAI, Germany) and Ira Haraldsen (Oslo University Hospital, Norway) showed how AI can power new research innovations aimed at improving dementia detection, diagnosis, prevention and treatment. Holger highlighted some of the challenges facing healthcare systems, many of which lack the capacity to screen, diagnose and adequately treat people with Alzheimer’s disease and dementia. Explaining that AI algorithms have the power to support more accurate, personalised diagnoses and detection, Holger outlined the goals of a newly-launched pan-European research project called PREDICTOM. This Innovative Health Initiative-funded project is creating an AI-enabled platform comprising screening and diagnostic tools that could be deployed in the community and in the primary care setting. Holger also spoke about the ADIS project, which is evaluating correlations between sleep disturbances and markers of neuroinflammation, using machine learning. Ira, speaking after Holger, focused her presentation on the AI-MIND project, a Horizon 2020-funded project which is developing tools for dementia risk screening in people with mild cognitive impairment (MCI). The AI-MIND tools, which are built around powerful machine learning algorithms, aim to shorten the diagnostic period from months or years to a matter of weeks. Ira explained how tools such as the AI-MIND Connector, which analyses data from functional brain networks measured using EEG and MRI, could provide access to a more personalised, rapid diagnosis delivered in the GP’s office.
The next presentation was entitled “AI for rare disease diagnosis: a perspective for AD”, delivered by Marcelo Martinez Conti of Foundation 29. Foundation 29 was launched by Julian Isla, whose son has a rare disease called Dravet Syndrome. Julian and Marcelo, who both have a background in computer science and IT, wanted to accelerate and improve the diagnostic process, and empower patients and their families to use, understand and share their clinical data. Foundation 29 therefore set out to develop AI-powered tools for doctors and patients, to support clinical decision-making and patient-driven diagnosis. Marcelo presented the dxGPT tool, which uses the ChatGPT system of large language models to suggest possible diagnoses when provided with information about symptoms. On the data side, the nav29 prototype tool is a personal health platform for patients to consolidate and share their medical records, ask questions about their health, and access summaries of patient-doctor interactions.
The next speakers, Soraya Moradi Bachiller and Daphné Lamirel (Public Involvement Officers at Alzheimer Europe), offered the perspectives of people living with dementia and their carers on AI. These perspectives were drawn from public involvement (PI) consultations in the context of research projects such as ADIS, eBRAIN-Health, AI-MIND and PatternCog. Soraya and Daphné identified some common themes in the responses, such as data privacy and confidentiality, bias and discrimination, and the impact of AI on the patient-doctor relationship. Explainability was also highlighted as an important consideration for people with dementia and their carers, including clear accountability for decisions made using AI-powered tools – coupled with access to tailored post-diagnostic support with actionable outcomes for patients.
The final speaker of the day, Saila Rinne, highlighted a few recent AI policy developments, including the new AI Act proposal, and the launch of AI and brain health research programmes by the European Commission. The AI Act proposal, for which provisional agreement was reached in December 2023, will categorise and regulate AI based on its risks to cause harm to individuals and society. Specific rules for high-risk AI aim to protect citizens from potential harm, whilst regulatory sandboxes are intended to promote innovation in a safe space. Turning to research and innovation, Saila described a new European partnership for brain health, bringing together the JPND, NEURON, and the Human Brain Project; and the Virtual Human Twins initiative, which aims to improve access to personalised prevention, early diagnosis, and tailored clinical pathways for all EU citizens. Concluding, Saila emphasised that patients and the public are the key stakeholders in these initiatives, explaining the EU approach to involving the public through advisory boards and representatives with lived experience of disease.