Title: Data Science for Health – Concepts and methods for learning on temporal health-related data

Medical research and medical decision support increasingly rely on data sciences advances. The research domain that encompasses advances on learning for health is vast and fragmented. In this tutorial, we will discuss following issues:

1. Forms of medical data used in medical research vs medical decision support

2. Learning on multidimensional clinical data – prediction and phenotyping

3. Taking time into account during learning on clinical data

4. Time and Ecological Momentary Assessments (EMA) – learning on mHealth data

This tutorial is for data scientists who want to apply and extend their methods for tasks in the medical domain. The focus is less on elaborate learning algorithms and more on the tasks of specifying the learning problems, preparing the data and making assumptions before applying simple and elaborate learning algorithms. Some of the approaches we see have been applied on public domain data, but most of them have been designed for tasks in concrete clinical or public health settings.