Hospitals use robots to perform simple tasks, such as delivering medications and supplies, or greeting people in the reception area. A team from MIT’s Computer Science and Artificial Intelligence Lab (CSAIL), however, thinks robots might be most effective by helping humans perform a more complex task: scheduling in labor and delivery units.
The team demonstrated how a NAO robot from Softbank Robotics assisted nurses in the labor ward at Beth Israel Deaconess Medical Center in Boston, making recommendations on where to move patients and even which nurse to assign to a C-section. This job is typically performed by a “resource nurse” and is crucial to maternity wards running smoothly. At Beth Israel, the resource nurse coordinates 10 nurses, 20 patients and 20 rooms at the same time
As MIT points out, labor wards are tough places for scheduling as nurses have to predict when a woman will arrive in labor, how long labor will take, and which patients will become sick enough to require C-sections or other procedures.
“The aim of the work was to develop artificial intelligence that can learn from people about how the labor and delivery unit works, so that robots can better anticipate how to be helpful or when to stay out of the way – and maybe even help by collaborating in making challenging decisions,” says MIT professor Julie Shah, the senior author on a paper on the topic.
How the robot works
The robot was trained by watching humans perform the tasks – this is called “learning from demonstration” in robotics circles. However, this technique had never been applied to scheduling before. To overcome this, the team trained the NAO robot to look at several actions that human schedulers make, and compare them to all the possible actions that are not made at each of those moments. From there, it developed a scheduling policy that can respond to situations that it has not seen before.
“Rather than considering actions in isolation of each other, we crafted a model that understands why one action is better than the alternatives,” says Shah. “By considering all such comparisons, you can learn to recommend which action will be most helpful.”
The policy is “model-free,” meaning the nurses do not have to train the robot by manually ranking each possible action in each possible scenario.
Robot’s success rate
Nurses accepted the NAO robot’s recommendations 90 percent of the time, MIT says. The team also demonstrated that human subjects weren’t just blindly accepting advice – the robot delivered consciously bad feedback that was rejected at the same rate of 90 percent, showing that the system was trained to distinguish between good and bad recommendations.
Nurses had almost uniformly positive feedback about the robot. One said that it would “allow for a more even dispersion of workload,” while another said that it would be particularly helpful for “new nurses [who] may not understand the constraints and complexities of the role.”
Shah says this new techniques have many uses, from turning robots into better collaborators to helping train new nurses, but the goal is not to develop robots that fully make decisions on their own.
“These initial results show there is tremendous potential for machines to collaborate with us in rich ways that will enhance many sectors of the economy,” says Shah. “The awkward robots of the past will be replaced by valued team members.”