Socially assistive robotics: robot-assisted diagnosis of women’s depression around childbirth
Main Supervisor: Ginevra Castellano
Assistant Supervisor: Fotis Papadopoulos
What is your educational background?
Master of Science in IT and Cognition, University of Copenhagen
Bachelor of Engineering in Mechanical Engineering South China University of Technology
Why did you apply to WOMHER's interdisciplinary graduate school? This is a great chance for me to work in an interdisciplinary environment, which gives me the opportunity to communicate with colleagues with the same interests but different expertise.
Tell us more about your research project?
There are 16,000 women who suffer from peripartum depression (PPD) in Sweden every year and the majority remain undiagnosed and untreated. This research project aims to design, develop and evaluate new methods for social robot-assisted diagnosis of PPD in women.
What is a social robot?
A social robot has a physical body that has certain human characteristics (eg head, eyes, mouth, arms) that allow it to interact with humans in ways similar to how humans communicate and interact with the world (e.g. through speaking or using social signals such as facial expressions or gestures).
In this project, we will develop software programs that together with sensors such as cameras and microphones make it possible for a social robot to learn to interact with humans and to conduct a psychiatric interview and at a later stage to be able to calculate a probability if a woman has PPD or not. There are currently no studies examining robot-assisted diagnosis of depression. It is therefore important to conduct research to understand how to involve clinics and patients in the development process for these new tools.
The project will address several important scientific challenges in socially assistive robotics for achieving user-centred robot-assisted diagnosis of peripartum depression (PPD). The first scientific challenge is how to design an interview scenario that enables robots to assist clinicians in diagnosing PPD. The second scientific challenge involves how to develop machine learning-based methods for the early identification of women at high risk for PPD and for robot-assisted diagnosis that enable a robot to learn from clinicians how to conduct interviews and automatically predict PPD types. Finally, the third scientific challenge is how to develop methods for robot-assisted diagnosis that are socially accepted by patients and clinicians and that complies with principles of trustworthy AI.
What do you hope the impact of this project to be?
Results of our project can lead to clinical systems for diagnostic support in primary care and/or psychiatry.
Other information, references and links
Zhong, M., Bilal, A. M., Papadopoulos, F. C., & Castellano, G.. (2021). Psychiatrists’ Views on Robot-Assisted Diagnostics of Peripartum Depression. In Advanced Data Mining and Applications (pp. 464–474). Advanced Data Mining and Applications. https://doi.org/10.1007/978-3-030-90525-5_40
Uppsala social robotics Lab: https://usr-lab.github.io/
PhD student at Department of Information Technology, Vi3; Human Machine Interaction