Description

This course offers a project-oriented, multidisciplinary introduction to the basics of data driven medicine. Orientation, fundamental concepts, and methodological approaches are provided by lectures. In addition, the participants will also form small interdisciplinary teams including students of computer science as well as medical students in order to plan and implement an own project, which targets prediction or decision support generated from medical data.

Data play an important role in medicine: Intensive care relies on monitors presenting and analysing real-time patient data, medical imaging has become a domain of massive data processing, diagnostics rely on laboratory data, and the importance of data is ever increasing: Wearable sensors, mobile communication devices and respective apps will produce data streams, which support preventive measures in healthy individuals or allow screening as a basis for data-based prevention of diseases. Last but not least: molecular biology (e.g. by gene sequencing and gene expression analysis) introduces new biomarkers, which enable new minimally-invasive diagnostics and approaches to tailoring treatments based on individual characteristics of patients (precision medicine) – which would never be possible without sophisticated processing of huge amounts of data.

Medical decision making in general will be markedly influenced by data processing and data analytics. Thus, we can expect data driven medicine to gain momentum in the nearer future. After the course the participants should be able to

• extract, load, transform data from relevant medical data sources of medical data via application programming interfaces
• consider the role of natural language in medical documentation
• access and use medical terminology servers
• transform given medical data to standardized representation formats (RDF triplets, i2b2-star-schema)
• adopt basic methods of natural language processing for providing semantic enrichment and text mining
• access, design, provide and use medical metadata repositories and ontologies
• assess data quality of medical data sources
• adopt basic methods of bad data curation
• use data to predict important outcomes (predictive modeling)
• apply and validate basic machine learning algorithms to medical data