SIS Professor Dania Bilal Works with PhD Student & Others to Learn How to Use Social Robots to Help People with Dementia
SIS Professor Dania Bilal has explored research in recent years centered on artificial intelligence (AI) from an information sciences perspective, exploring how people interact with various AI and robot interfaces. She most recently partnered with doctoral student, Fengpei Yuan, and her advisor, Professor Xiaopeng Zhao, at the University of Tennessee Department of Mechanical, Aerospace, and Biomedical Engineering, and Assistant Professor Ran Zhang at the Miami University Department of Electrical and Computer Engineering, to study how social robots could provide certain therapies for people with dementia to mitigate cognitive decline.
Yuan is the first author for the paper that resulted from this research, “Learning-Based Strategy Design for Robot-Assisted Reminiscence Therapy Based on a Developed Model for People with Dementia.” The paper was nominated as a Best Student Paper Finalist by the 13th International Conference on Social Robotics (ICSR 2021) in November 2021.
Yuan first met Bilal when she registered to take Bilal’s class Seminar in Human-Computer Interaction (HCI) and User Experience (UX) in 2018. The semester-long assignment in the class was to work on a research proposal, and for her project, Yuan chose a proposal about using robots to facilitate reminiscence therapy for people with dementia. Reminiscence therapy uses videos and pictures followed by questions designed to encourage clients to talk as much as possible. A social robot is one that can convey human-like emotions via its expressions, voice, and body language.
“We want to develop the robot so that the robot can adopt flexibility and talk to a person with dementia, in the context of reminiscent therapy,” Yuan said. “With this therapy, the more we engage the patient to talk, the more helpful and meaningful it will be to the patient. The social robot can have body movements and can talk and also has a tablet in its chest, so we think that it’s a good idea that we translate this reminiscence therapy to this robot.”
Bilal said that this is a new model of training the robot so it can actually attend to the emotional and cognitive behavior of the people it observes.
“That’s why it’s flexible, it also learns and provides the therapy that is needed according to the state of the patient, to their behavior and emotions,” Bilal said.
If successful, using social robots for reminiscence therapy could fill the gaps in a field that has a dearth of therapists and providers. Yuan said this work is particularly vital because the number of people suffering from Alzheimer-related dementia is increasing rapidly, with the current population numbering around 6.2 million.
“AI-enabled social robots may help to reduce the cost of healthcare and help to address the disparity issues by increasing the accessibility of high-quality care and treatment,” Yuan said.
Bilal has an interest in research focusing on older adults. When she read the proposal by Yuan she thought it was novel and interesting. Zhao, Yuan’s advisor, and Yuan reached out to Bilal to expand Yuan’s proposal and submit it for external funding.
While it may seem like science fiction to some to imagine robots providing therapy for humans, Bilal said people are becoming more accustomed to robot assistance in their daily lives and now is the time for innovative research such as this.
“Robots are expensive but competition will bring the prices down. So exploring the technology and doing research on human robot interaction becomes highly important,” Bilal said. “We want robots to be very positive and to make a difference in people’s lives.”
Using social robots and training them how to react to human emotions can go a long way to ensure people feel more comfortable getting help from robots, and it speaks to what is coming in the next wave of technology.
“This work makes contributions to promote personalization and behavior adaptation in the applications of socially assistive robots to technology-empowered healthcare,” Zhou said.
The research itself consisted of collected data from a simulation using the social robot; the researchers wanted to ensure the robot will be ready for real human interaction before taking the big leap of conducting human-centered studies, Yuan said. This initial research is focused on creating a dynamic model that would program a robot how to react to an individual’s emotions and how to answer questions within the context of dementia patients.
“Using robots to automate the RT treatment with deliberately designed reinforcement learning algorithms well matches the current tide of using AI to improve people’s life,” explained Zhang.
Zhang created an algorithm that teaches the robot to learn as it works so it can act based on other’s reactions. For example, the robot would be able to sense human emotions such as anger, discomfort, or frustration from verbal and non-verbal cues.
“The more research we do, the more acceptable the idea of using robots is going to be, so we definitely need to do lots more work,” Bilal said.