The objective of this PhD thesis is to propose models that predict and explain learner performance on the basis of their interaction traces from the 3D software. The PhD student will be required to:

• Design a trace model that represents learner activity on the application and defines the set of elements to be observed.

• Align these elements with data from other collection sources (behavioral - eye tracking, physiological – electrodermal activity, psychometric - questionnaire responses), interpreted by experts in human learning and anatomy to build a multidisciplinary trace.

• Analyze this trace using indicator calculations and supervised and unsupervised learning methods to define learner performance prediction and explanation models, as well as a set of personas representing typical learner profiles.

Once defined, the personas and performance indicators will be presented and discussed with experts in the field of human learning. This will enable them to design adapted pedagogical scenarios, which will then be integrated into the 3D learning tool.