Course description
Over the last five to ten years machine learning (ML) methods has gained widespread use with both sports and health data. ML methods can be used with both accelerometry or heart rate data for health or sports purposes or for simple clinical studies to find important patterns in the data. The possibilities seem almost endless. An important strength of the ML methods is that it can model highly complex data, which is common attribute of most sports and health data. However, the introduction of engines like the ChatGPT or Bard also suggests that understanding the strengths and weaknesses of this branch of statistical methods is important to disseminate quality health and physiological information from the sports and health data.
Expected learning outcomes
The present course will cover both unsupervised and supervised learning. Within unsupervised methods the focus is on principal component analysis and clustering, whereas with supervised methods we will cover both classification and numerical prediction using methods like decision trees and neural networks. The course will be highly practical, and you will get hands-on experience with model selection, learning, tuning and evaluating performance and generalizability. All work will be done using R. Prior experience in R is not required but would facilitate learning experience.
Teaching methods
A mixture of lectures, group assignments, workshops and student presentations
Price
The course is free of charge for PhD students enrolled at the Faculty of Health Sciences at SDU and for PhD students enrolled at other universities that have joined the Open Market agreement and/ or the NorDoc agreement.
The course fee for other participants is:
DKK 2.700
EUR 361