MIT robot helps deliver babies
NEW YORK — Would you trust a robot to help deliver your baby?
Robots could eventually play integral roles in labor wards, according to findings from MIT’s Computer Science and Artificial Intelligence Lab Robots.
Robots are currently employed in hospitals to carry out simple actions, like dispensing medication. But can they understand patient needs and make scheduling decisions?
The researchers have been working for the past two years to determine whether robots can be more than just helpful companions.
They’ve been conducting experiments to see if a robot can serve as an effective “resource nurse.” That’s the nurse in the labor and delivery unit that’s in charge of assigning other nurses to care for patients.
“It is really one person making these decisions. It’s a very complex environment and a very hard job,” said MIT professor Julie Shah, the senior author of the study. The job requires effective decision making — which room should a patient be assigned to? Which nurse should perform a C-section? — in a fast-paced, often unpredictable environment.
To conduct the study, the researchers trained a Nao robot to learn from nurses’ scheduling decisions and to understand why they made those decisions opposed to the alternatives.
The robot takes into account the complexity of patients assigned to particular nurses, break schedules and more. It can also determine nurses’ availability on the floor.
The researchers then had the robot make suggestions for doctors and nurses in Beth Israel Deaconess Medical Center in Boston. At Beth Israel, the resource nurse coordinates 10 nurses, 20 patients and 20 rooms at the same time, according to the study’s authors.
90% of the time, the Nao robot made suggestions that doctors and nurses carried out, the study found.
According to Shah, there are two important use cases for the technology. It can be an effective training tool for novice nurses and can help make better decisions in the labor ward.
The researchers performed the experiment first with a computer-based support and then with a robot to compare whether nurses complied differently with the suggestions of robots versus computers.
They didn’t see an over-reliance on the robot, according to Shah, who also led a similar study that applies the same system to missile-defense scenarios.
Next up for the labor robots, Shah said, is to expand the research to other labor units in more hospitals.
“We are looking to scale this, but this is a safety critical domain so we’re making sure we take measured steps,” Shah added.