Not Sure if Your Baby Is Hurt or Just Hungry? Let the Robot Decide

Can you tell which of these babies are in pain and which might just have a bad case of gas? According to Dr. Brahnam of Missouri State University, her new software system can. According to her abstract, the best algorithm they found determined correctly 90.20% of the time that their guinea-baby was in fact being lanced in the foot and not one of 3 other "nonpain stressors".

Using this technology 9 out of 10 babies in pain will be referred to the human maternity ward attendants (who will inevitably be out in the smoker's hut, using the robo-nanny as a crutch to avoid responsibility and the wrath of new mothers). If the robot selectively chooses which children will be the unlucky number 10, then we'll never know.

To their credit, researchers are following up this study with video studies to get more accurate results. Also, the technology may be useful in determining when patients have woken up from anesthesia, but can't scream. The only danger there is if the AI develops a curiosity as to what the pain looks like. This is an entirely feasible possibility if the system is designed to refine its own pattern recognition algorithms.

Thanks to MIT Technology Review

No comments: