Using socially assistive robots, pain researchers are looking to relieve distress for children during painful hospital procedures.
Editor’s note: During the International Symposium on Paediatric Pain (ISPP), five pain researchers participated in the PRF-ISPP 2023 Correspondents Program – made possible by generous contributions from Solutions for Kids in Pain (SKIP) and the Centre for Pediatric Pain Research (CPPR). As we prepare for ISPP 2025 – taking place 17-20 June 2025 in Glasgow, UK – we’re taking a look back at some highlights of ISPP 2023, and some of the people who made them possible.
Drs. Samina Ali (University of Alberta, Canada), Mary Ellen Foster (University of Glasgow, Scotland, UK), and Frauke Zeller (University of Edinburgh, Scotland, UK) participated in a symposium at ISPP 2023. Throughout their session, covered below, they shared their experiences working on a multidisciplinary international team utilizing a “user-centered co-design” to create a new robot system to relieve distress for children, and discussed the ethical implications of using AI in healthcare.
Distress in hospitalized children seems unavoidable. The negative effects that clinical distress can have on children in the short and long term are well known, and may include fear and inability to accept procedures, a phobia of interventions such as intravenous (IV) insertions, or avoidance/overuse of healthcare services. To help address this issue, Samina Ali (University of Alberta, Canada), Mary Ellen Foster (University of Glasgow, Scotland, UK), Frauke Zeller (University of Edinburgh, Scotland, UK), Jennifer Stinson (The Hospital for Sick Children, Toronto, Canada), Ron Petrick (Heriot-Watt University, Edinburgh, Scotland, UK), David Harris Smith (McMaster University, Canada), and Sasha Litwin (The Hospital for Sick Children, Toronto, Canada) have formed an international collaborative group to employ an approach called socially assistive robotics (SAR).
SAR, initially defined by Feil-Seifer and Matarić in 2005, focuses on providing user assistance through social rather than physical interaction. These robots have been used across populations of all ages in tasks related to tutoring, physical therapy, daily life, and emotional expression. The National Science Foundation (NSF) in the US has even established the NSF Expedition in Computing – a collaboration among Yale University, the University of Southern California, Stanford University, and the Massachusetts Institute of Technology – and funds initiatives focused on advancing SAR technology.
Ali, Foster, and Zeller’s plenary session on the topic asserted that some of the problems facing current health systems are too complex to be resolved by one profession, and concluded that interdisciplinary research and collaboration are the best methods of achieving solutions. Doing so requires an awareness of the environment and the times in which we live, and tailoring of our approaches based on individual situations. For example, because children are generally receptive to technological solutions, SAR have particular promise in the field of children’s health. However, as the team highlighted, not all robots respond to children in real time, and many follow a script, thus limiting the extent of influence that SAR can have in assisting care. Children have grown accustomed to various types of technological devices, such as smartphones and their virtual assistants; therefore, a more interactive and immersive approach may be more efficient. Artificial intelligence (AI) could allow for even more autonomy in these robots and improve the overall experience of patients and their families.
The system was based on an NAO robot and has been designed to be used in the children’s emergency room during the insertion of an IV. The device is used as a way to support child engagement and to better understand their visual cues. It would recognize how much the child was focusing on the IV insertion and attempt to distract them with a dance or other techniques. The robot could also explain to children what to expect during the day in the hospital, employing a hopefully more friendly and acceptable approach than would a non-familiar adult.
The process of creating and implementing this AI-enhanced social robot was quite complex. Robots need very strict guidelines and commands in order to react appropriately to external stimuli. As they developed the concrete technical specifications, the team considered questions such as: How can the robot differentiate if the facial expression of the child means discomfort or simple boredom? How can SAR enhanced by AI detect non-verbal cues in the child and know what action to take? In a published article from the collaborative group, the researchers proposed using AI-powered image recognition and natural language processing to assess human actions and emotions that will address the questions that were raised.
There are also ethical considerations to SAR implementation such as avoiding bias in its design. For example, the benefits of SAR in marginalized children could be compromised if the unconscious biases of those programming the machine are translated into its code. Similarly, some racial prejudices may be reinforced if the robot does not address diverse language, accents, and physical appearances, and there needs to be a design that accommodates children with disabilities or atypical/neurodiverse behaviors.
To develop this AI-enhanced robot that considered all of these factors was not simple, as it required thorough analysis of ethical concerns and technical implementation. Recent work has suggested that the addition of AI may help the robot assess the child’s developmental age – which may not match their chronological age in some patients. There are also efforts being made to adapt the SAR to other languages and cultural contexts.
While discussing the programming of the robot, the team also noted the importance of real-world considerations, such as aesthetics. What does the robot look like? How does its appearance impact patients, especially when working with children? What physical appearance would optimize children’s receptivity?
A robot that looked “frightening” to the eyes of children would affect its purpose in children’s healthcare clinical practice. Additionally, the necessity of a “kill switch” – a means to immediately suspend the robot’s operation – was also discussed. For example, if the patient was to experience a seizure, a robot who is dancing and talking to the patient would need to be immediately shut down. This quick response must also be easily accessed for a litany of additional safety reasons, and has been incorporated in the design of the robot.
In brief, this innovative approach presented by Ali, Foster, and Zeller has the potential to greatly benefit children’s healthcare, and a randomized clinical trial of the technology has recently been completed. Based on results from previous similar studies, which found that children were receptive to the AI-enhanced robot during their IV insertion and that the robot led to a modest positive impact on child – and even parental – distress, the team expects to find similar positive effects from the AI-enhanced robot developed in this project. It will be exciting to track this technology as it progresses.
Jéssica Ding is a master’s student at McGill University, Montreal, Canada.