Crowded hospital emergency departments (EDs) can frustrate waiting patients and overwhelmed medical staff. If you’ve visited an emergency department in the last year or two, you likely experienced crowded conditions and long wait times. Earlier this year, the CDC reported that the percentage of non-urgent and semi-urgent ED visits increased from 2015 to 2016. The report mentioned a release on the latest ED trends by the American College of Emergency Physicians (ACEP) that highlighted the effect of health insurance policies that deny coverage for non-emergency care on the increase in non-urgent ED visits. Earlier this year we wrote about MobileSmith’s paper on the effective use of mobile apps to reduce ED crowding.

Professor Goldie Nejat and a team of engineers from University of Toronto’s Autonomous Systems and Biomechatronics Lab and Institute for Robotics & Mechatronics (IRM) are developing a potential solution to ease wait times in Canadian hospital EDs. Nejat’s solution combines robotics and machine intelligence. The group is currently building algorithms to control the interactions between two semi-humanoid, four-foot tall robots and ED visitors. The robots are off-the-shelf “Pepper” models from SoftBank Robotics; they can detect and respond to voices and body language in response to input from four microphones, two HD cameras, and a 3-D depth sensor.

The Peppers’ initial assignments in ED rooms will include intake data collection, patient monitoring, and companionship. Acting as an intermediary between ED patients and healthcare professionals, the robots gather information and prioritize patient need urgency according to the patient’s symptoms. The system then transmits the report to a nursing station. Nejat’s goal is to break the logjam that comes from collecting all the required information collection. Following patient admission, the robots could monitor their conditions, ready to report if something goes wrong. Each Pepper will also learn to provide companionship, a role that will require the robots sense how the patients are feeling along. Nejat labels the ability to determine how people are feeling as “emotional modeling.” For example, if a patient reports feeling fine but is hunched over with a furrowed brow and speaking with a strained voice, the robot has to be able to recognize the mismatch, respond to the patient appropriately, and know when it’s time to call in the medical cavalry.

The U.T. group plans to start testing the robots in a hospital emergency department within a year. Nejat foresees a time when robots could use influence strategies and techniques to promote social interaction and persuade them to comply with treatment plans and to take their medication. Using robots instead of healthcare workers could save money and potentially shorten ED wait times and improve the overall patient experience.