20 | wetenschapsdag 2023 Lopend onderzoek in de regio Amsterdam 4 Auteurs L. Rijken, K.K. Yeung, M.P. Schijven Abstract titel Developing Trustworthy Artificial Intelligence (AI)-driven Tools to Predict Abdominal Aortic Aneurysm Progression and the Risk of Adverse Cardiovascular Events: the VASCULAID-RETRO Study Background To date, it is unknown which patients with an abdominal aortic aneurysm (AAA) will suffer cardiovascular events or in which patients the AAA will progress. The VASCULAID-RETRO study aims to develop artificial intelligence (AI) algorithms able to evaluate the extent of AAA disease progression and risk of cardiovascular events. Methods The VASCULAID-RETRO study aims to leverage retrospectively collected data of at least 5000 AAA patients from multiple European clinical centers for the development of AI-algorithms. Initially, a robust data infrastructure network will be established to gather standardized data from all six participating clinical centers. After collection of imaging, -omics data, and (reported) clinical patient data, AI-tools will be developed using this data. Automatic anatomical segmentation on images and image analysis on US, CTA and MRI will be performed. Moreover, prediction algorithms for each data type (imaging, -omics, and clinical data) will be created separately. These prediction algorithms will be merged using fusion AI models to build a comprehensive prediction algorithm based on multi-source data to generate overall risk scores or probabilities for AAA progression and the risk of cardiovascular events. Results Ethical approval for retrospective patient data collection have been secured by all clinical partners. Currently, the data infrastructure for the collection of the retrospective data is being developed. Patient data from electronic patient files will be collected in Castor EDC and imaging will be stored on an XNAT server. Conclusion FUTURE PERSPECTIVE: The VASCULAID-RETRO AAA study is part of the VASCULAID project, an European Horizon-funded research project. Similar AI algorithms will be developed for patients with peripheral arterial disease (PAD) of the lower limbs. Following the VASCULAID-RETRO studies for AAA and PAD patients, prospective studies will be performed in which more data will be collected and the developed AI-algorithms will be validated for identifying AAA and PAD patients at high risk of disease progression and cardiovascular events.
RkJQdWJsaXNoZXIy MTk4NDMw