Sign In

Communications of the ACM

ACM TechNews

Does AI in Healthcare Need More Emotion?


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook

Emotion AI or affective computing has not been widely applied in healthcare, as can be seen in a recent literature review.

Credit: Getty Images

Emotion artificial intelligence (AI) or affective computing applications in healthcare are rare, with many scientists skeptical of the technology's utility.

A team at Spain's University of Seville reviewed 156 papers that used AI in pregnancy studies, to explore development of a mobile application or wearable for detecting health conditions.

The researchers determined AI has been used for detecting and monitoring fetal health, congenital birth defects, preterm birth risk, and gestational diabetes over the last 12 years, albeit with little clinical testing.

The University of Seville team only found two AI studies including emotional parameters in medical literature: one on the relationship between pre-eclampsia and anxiety, and one about patient reactions to prenatal tests.

Ohio State University's Aleix Martinez said AI's generally poor ability to interpret emotions reflects the same shortcomings in humans, while the University of Seville's Andreea Oprescu sees value in including emotional parameters for many healthcare areas, like pregnancy.

From IEEE Spectrum
View Full Article

 

Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA


 

No entries found