Healthcare @ SAP Innovation Center Potsdam Dominik Bertram 28. November 2013
Jan 01, 2016
© 2013 SAP AG. All rights reserved. 2Public
IS-H Analytics - Real-time analysis of hospital patient management dataAndrew McCormick-Smith, Christian Heller, Spyros Antonopoulos
“This would change the way I do my job.” - Charité University Hospital Berlin “ ”
Reports in < 1 second with in-memory, old system 55s
Challenge Current patient management systems are too slow for real-
time analysis, making what-if planning impossible
Solution Re-implemented a selection of IS-H reports in HANA Results delivered in sub-second response time New analytical applications can now help drive cost-savings
and more efficient resource allocation
Benefits Cost savings for hospitals Improved patient experience
Flexible, real-time analysis – no need for materialized aggregates
© 2013 SAP AG. All rights reserved. 3Public
Medical ExplorerDominik Bertram, Massimiliano Marcon, Gennadi Rabinovitch, Matthias Steinbrecher
Challenge Integrate multiple sources of patient data, including both text (e.g.
doctor’s letters) and structured data (e.g. tumor registry) Pick out patients whose treatment history satisfies very complex
criteria Compare metrics like survival times, quality of life, treatment
response across different patient cohorts
Achievements Generic medical data model makes it easy to combine data from
many different sources Intuitive web UI supports analysis of patient cohorts based on
customizable attributes Solution will go live at the National Center for Tumor Diseases
(NCT) in Heidelberg in late 2013 / early 2014
Unified access to multiple formerly disjoint data sources
© 2013 SAP AG. All rights reserved. 4Public
ProteomicsDBJoos-Hendrik Boese, Lars Butzmann, Dave Schikora
Collaboration project with HANA Platform Core Team Walldorf and Technische Universität München
Goals Provide a public data repository for storing and sharing proteomics
experiment data Support rapid calculations across the entire database
Achievements Went live on https://www.proteomicsdb.org on June 11th 2013 Currently holds over 3.5 TB of proteomics data covering over 90% of the
human proteome
© 2013 SAP AG. All rights reserved. 5Public
Proteome-based Cancer DiagnosticsJoos-Hendrik Boese, Siegfried Gessulat, Christopher Ozdoba
Challenge Proteome data from blood samples could provide minimally invasive
early-stage cancer diagnostics Very large data sets (160Mio data points/sample) need to be analyzed
to detect disease patterns and do diagnostics
Solution Implemented complete analysis pipeline for proteome-data in SAP
HANA Built an analysis modeling tool that allows researchers to manipulate
the detection pipeline interactively
Benefits Large studies now feasible to identify complex disease patterns Intuitive interaction with data for researchers and doctors
Intuitive interface for complex analysis pipeline
Large scale studieson high resolution data now possible
© 2013 SAP AG. All rights reserved. 6Public
Virtual Patient PlatformMorten Ernebjerg, Valentin Flunkert, and Marcus Krug
Challenge
• The Virtual Patient model is a predictive computer model which captures how a human cells works; it contains almost 2500 differential equations
• Virtual Patient simulations could offer a revolutionary way of finding the best treatment for each cancer patient….
• …but the processing time and data volume previously made large-scale use impossible.
Achievements
• Optimized model solver to run 5000 times faster - large-scale simulations can be done in hours instead of weeks
• Built a comprehensive data model in HANA for quick analysis
• Created web-applications for doctors and scientists to run simulations and analyze the results on HANA
5000x faster simulations
Real-time analysisof results data now possible