MIP: Advanced Data Processing and Analytics for Science and Medicine Full text

K. Filippopolitis, Y. Foufoulas, M. N. Garofalakis, A. Glenis, Y. E. Ioannidis, T.-M. Karampatsis, M.-O. Katsouli, E. Mailli, A. Papageorgiou-Mariglis, G. Papanikos, G. Pikramenos, J. Sakellariou, A. Simitsis, P. Ducouret, P. Ryvlin, M.-G. Spuhler
Abstract. We present the Medical Informatics Platform (MIP), an online collaborative platform for the scientific and medical community. It federates de-centralized patient data located in hospitals, helping clinicians, clinical scientists, and researchers identify patterns unique to diseases and provide clear diagnoses and personalized treatments. The platform enables users to access harmonized medical data from pre-processed neurophysiological and medical records, and research cohort datasets without the need to transfer original clinical data. This functionality facilitates exploration and analysis of medical data while preserving the privacy and security of sensitive patient information. The MIP blends data science and machine learning with data technology, and especially data integration, secure computation, decentralized distributed query execution, and low level, efficient execution of science pipelines exploiting features of modern data engines such as vectorization, parallelization, and JIT compilation. The MIP is the result of a multi-disciplinary, multi-year effort among computer scientists, clinical scientists and medical professionals. To date, it has been deployed and used in 40+ hospitals across Europe and another 12 installations are in process.