Seminar: Interactive Data Science in Digital Health (MSc, PhD)

Both data scientists and medical researchers such as epidemiologists conduct data-driven research to discover new knowledge and create evidence. Interestingly, the methodologies of both disciplines differ considerably, as you will learn in the seminar. At a glance, epidemiologists conduct carefully designed experiments to gather new data for downstream analysis, whereas data scientists exploit existing data, e.g., for visual data exploration purposes. Also, data visualization and interactive data analysis methods differ considerably in both domains.

How to evaluate new digital and mobile health applications? Which methods or study designs are most appropriate? What are legal and regulatory requirements? How can data quality problems be addressed? How to cope with the complexity of data? And when and how should users be involved in the development and evaluation process?

This two-day block course will address these and more questions from the viewpoints of clinical research methods on the one hand and interactive data science on the other hand. Together, we build bridges between a) experimental methods and concepts for evaluating medical health and b) statistical and machine learning tools and interactive data analysis methods. The focus will be on concepts, study planning, and the choice of analytic designs and methods.

The course will not be mathematical. Despite this, students should possess a basic understanding of data science tools such as statistical methods (e.g. linear regression). In the course, we plan for practical group exercises where students design a mobile health study.

In the weeks after the two-day block course, students will conduct (continue) individual project work according to real-world application examples in digital health. The project results will be submitted to the teacher in form of a written document, according to a document structure pre-defined by the teacher.

At the end of the seminar, students have gained a deeper and broader understanding of data-driven decision-making in health applications.

In particular, participants should have an awareness for the critical points in study design and conduct, and for the data analysis workflow.

Given this knowledge, students will be able to critically appraise protocols and publications

1. What is digital and mobile health (mHealth)?
2. How do epidemiologists and data scientists think?
3. Experimental methods and concepts for evaluating mHealth
4. Data scientists’ approaches to analyzing (non-experimental) data
5. Overview over ethical, legal and regulatory frameworks for mHealth
6. User Engagement: What are the needs and wishes of end users?
7. Data, Tasks, Users: Strategies to plan and execute mHealth evaluations

Lectures and practical exercises (+data analysis task for Master students)

• Active participation in the block course, including contributions to discussions and feedback
• Quality of project results
• Quality of written project result documentation