Univ.-Prof. Dr. rer. nat. Andreas Schuppert

Direktor des Instituts für Computational Biomedicine der RWTH Aachen.

Themengebiete: Digitale Patienten, Hybride Modellierung durch Kombination von AI und mechanistischen Modellen, Explainable Artificial Intelligence, Anwendung von neuen Computertechnologien wie Quantencomputing für biomedizinische Fragestellungen

Was Sie in meinem Modul: Data Analytics II – Predictive Analytics & Machine Learning erwartet:

The module will provide a comprehensive overview on the machine learning methods and their applications in medicine. The students should learn the specific features of the most used methods as well as their caveats and workflows for validation of machine learning results.

The modul will introduce to the modern methods in machine learning and data mining with a special focus on predictive modelling in medical data environments. It will provide an overview on the standard methods both in unsupervised and supervized learning with a focus on “fat data” analytics, e.g. for –omics data, as well as ‘lean data’ analytics as arising in observational patient data. Special emphasis will be laid on the assessment of the caveats arising from high dimensionality (“curse of dimensionality”), missing data and hidden parameters. To tackle the pitfalls of machine learning in medicine, the course will train the standard validation methods used for quality assessment of pattern recognition. The course will provide lectures for theoretical foundations and experimental hands on training.

Warum ist das Modul für Medical Data Scientists wichtig?

Artificial Intelligence, especially Machine Learning, has the potential to transform the medicine of the future. It enables to extract structured information from huge data sets which become available in the medical context and to create predictive models enabling to predict the outcome of complex therapies. Hence, transforming highly complex, heterogeneous data to tangible information and predictions, these technologies offer doctors a new tool towards optimal interpretation of diagnostic data and data based decision support, even in highly complex settings.

To realize these promises a good understanding of the concepts of machine learning and the multitude of tools, not to forget its validation, is essential in order to design and apply the optimal computational solutions for the specific medical purpose.