项目介绍

2026年4月24日

What are you going to do?

The central aim of this project is to contribute to the data-analytic goals of the Stress in Action programme by comparing and integrating machine learning (ML) approaches with dynamic intensive longitudinal data (DILD) modelling. As part of the Data Analytic Support Core (DASC), this project focuses on how the dynamic interaction between contextual stress exposure and multicomponent stress responses in daily life can be best modelled, while accounting for individual differences.

A key innovation is the deliberate combination of ML and time-series modelling techniques. Whereas ML methods are typically optimized for data reduction and prediction, and DILD approaches emphasize explanation and interpretability of temporal processes, their integration is expected to yield both accurate predictions and a deeper understanding of what these predictions represent in daily life.

A central focus of the project is the role of qualitative data. Qualitative information will be used to contextualize and interpret quantitative patterns, helping to clarify how participants experience, interpret, and respond to stress in daily life. In addition, qualitative insights will inform model development by identifying meaningful features and improving the ecological validity of predictions.

Finally, the project will examine how much data is required to obtain reliable and meaningful predictions. This involves combining quantitative evaluations of model performance with qualitative approaches to understand, in practice, how many data points are needed to make valid inferences about stress processes in everyday life.