A study carried out by scientists from Weill Cornell Medicine (New York, United States) demonstrates that an algorithm is able to know which embryos fertilized in vitro will be more likely to originate pregnancies that end satisfactorily. The researchers, who have published their work on the NPJ Digital Medicine portal, they were based on 12,000 photographs, to which, one by one, they granted a rating according to their appearance.
A subsequent statistical analysis established a correlation between these factors and success in fertilization. The algorithm they designed, called Stork, got a 97% accuracy. For one of the authors, Dr. Olivier Elemento, director of the Caryl and Israel Englander Institute for Precision Medicine at Weill Cornell Medicine, this innovation "maximizes the chances of patients having a healthy pregnancy."
A group of European colleagues, in particular, from Italian and Finnish universities and hospitals, have completed another experiment that, thanks to technology, Detects development problems in newborns. His technique consists of discovering patterns of movement that could be linked to serious ailments, such as cerebral palsy. In their conclusions, disseminated through the Acedia paediatrica magazine, highlights the role of another algorithm.
From recordings of video in three dimensions and mathematical calculations, it can be ascertained if the motor maturity of a baby corresponds to its real age. The margin for moving forward is as wide as the controversy that some of these solutions raise. For example, in a collaboration between Chinese and North American experts, collected by the prestigious Nature Medicine, the potential of artificial intelligence for child diagnoses is discussed.
Concerns like this have led to Stefania Druga, from the Massachusetts Institute of Technology (MIT, United States), to deepen the interactions between minors and automatic devices. With this purpose he has created the Cognimates platform, for the kids to program and personalize your digital tools and even your robots. And so, almost without realizing it, they end up becoming masters of machine learning. Druga argues that, in this way, boys can make "more informed and critical use of technology".
Like many of his colleagues, Druga believes that artificial intelligence has a lot to learn from children. Without going further, the psychologist Alison Gopnik remember that the minds of the little ones can provide clues to scientists and computer scientists about how to improve the techniques of machine learning
. "Even one-year-olds can be flexible and make generalizations, but the machines themselves find it very difficult," he concludes.