Alzheimer's disease (AD) is a neurodegenerative disorder leading to the development of dementia, and a major cause of morbidity and mortality in the ageing population. In the IHI-PROMINENT project we will develop prediction models of AD diagnosis and disease progression, based on existing data from Swedish and European health care databases, registries and cohorts. The Swedish Dementia Registry is one of the largest AD databases in the world, with over 100,000 patients followed longitudinally. This registry has been linked to health care resource utilization data, mortality, biomarker and other data. We aim to use the high dimensionality of this dataset to develop novel prediction models of AD diagnosis, and to identify patients at high risk of disease progression. This is a highly relevant topic as new disease-modifying therapies for AD are being introduced, the first two drugs have recently received regulatory approval in the US and are currently under review by the European Medicines Agency (EMA). IHI-PROMINENT started in 2023 and will continue until 2028. Over the past year, using computational resources at NAISS, we have developed a novel deep learning model for simulating health care utilization data and predicting disease trajectories and care needs in Alzheimer’s disease. The model is using a transformer-based architecture and was trained on tokenized health care data from the Swedish Dementia Registry combined with other national health care datasets. The accuracy in validation datasets reached >40% top-1 correct predictions and >80% correct top-5 predictions. The model shows promise as a general tool for describing disease progression, projecting care needs and determining the potential cost-effectiveness of novel disease-modifying therapies in Alzheimer’s disease. Continued work is needed to further train and validate the model and develop use-cases for application. We are applying for a continuation project for this purpose.