Not all children are as happy as we usually expect. About 20 percent suffer from anxiety and depression, but those internalized disorders are hard to spot.

A study published on Jan 16 in the journal PLOS ONE describes an AI-enabled tool to screen children for internalized disorders early and accurately.

Researchers from the University of Vermont and the University of Michigan combined a sensor and an algorithm with the method that picks up on the children's behaviors and feelings, such as anxiety.

The children in a dimly lit room were told to look at a covered glass box that contained a fake snake. Then the researchers scored their responses, traditionally through the recorded video, and through a wearable motion sensor with a machine-learning algorithm.

The new tool identified differences between children with internalized disorders and those without. The accuracy was 81 percent, which was better than the standard parent questionnaire, according to the study.

The algorithm learned that children's movements before the snake was revealed were the most indicative since those with internalized disorders tended to turn away from the potential threat.

It showed that they had more anxiety, and the turning-away behavior was a negative reaction.

Just 20 seconds of data from the anticipation phase provided by the sensors and the algorithm was enough to make a decision compared with the traditional video-coding method that could take several months.

The new method opens up possibilities of large-scale screening to identify anxious and depressed children, says the study's co-author, Maria Muzik, at the University of Michigan, adding that early intervention is key because young children's brains are extremely malleable and respond well to treatment.