Journal Club GENOMICS AND DIGITAL HEALTH 2022-2023
Human-Centric AI Research in Service of Quality of Life
Professor Katarzyna Wac, PhD, is a Full Professor of Computer Science at the University of Geneva in Switzerland. She is also an Invited Professor of Health Informatics at the Department of Computer Science (DIKU) at the University of Copenhagen in Denmark, and has been affiliated with Stanford University and the Stanford Medical Centre since 2013. Professor Wac leads the Quality of Life Technologies lab, which aims to responsibly leverage daily life data sources to improve the quality of life of all individuals. The QoL lab's research interests include the fundamental and algorithmic problems, as well as the human-centric challenges related to the assessment and improvement of human behavior, well-being, health, disease self-management, and quality of life as it unfolds naturally over time and in context.
Disease patterns are changing, from accidents to slow and debilitating chronic conditions caused by harmful repetitive behaviors like poor sleep and nutrition. These behaviors affect physical, psychological, social, and environmental domains and have long-term impacts on individual’s life quality. Meanwhile, miniaturized mobile and wireless technologies in smartphones and wearables allow for minimally obtrusive assessments of behaviors and health risks, as well as modeling of long-term quality of life.
The Quality of Life lab uses advanced modeling and a reliable, privacy-preserving research infrastructure to capture "small personal data" from individual’s smartphones and wearables in a "Living Lab" approach. The chosen modeling approach integrates a mix of theoretical components from computer science (including latest deep learning-based, and other artificial intelligence approaches) and theories and models from social and behavioral sciences and preventive medicine. This talk will provide foundational concepts and practical realities, including human factors important in the design, development, and evaluation of digital biomarkers for life quality assessment and improvement in the long term. Furthermore, this presentation will discuss our recent research findings [1], which utilize machine learning techniques to transform an individual's observational behavioral time series into simulations for within-individual randomization of a set of interventions. This enables us to determine the causality between interventions and behaviors. We will also discuss [2] our use of transformer-based models to predict individuals' behavior related to their "sleep debt".