Artificial intelligence, thanks above all to algorithms that learn automatically (machine learning), is also increasingly entering the medical sector dedicated to mental disorders thanks mainly to the fact that human speech can offer a considerable level of detail and data that can be analyzed. This is the case of new research published in the Journal of Biomedical and Health Informatics according to which, with a machine learning algorithm, it is also possible to detect the signs of depression in children by analyzing speech patterns.
About one in five children suffer from anxiety and depression, disorders also known as “internalizing disorders.” However, unlike adults, it is more difficult for doctors to diagnose a disorder such as depression because children cannot reliably articulate their emotional suffering. This difficulty then gives rise to a series of problems that can lead some parents to give up treatments that could be very important.
It is thought that for most depressed children this state is not diagnosed or treated, as reported by Ellen McGinnis, a psychologist at the University of Vermont and one of the authors of the study. This study has seen the researchers use an automatic learning algorithm to analyze the main characteristics, also on a statistical level, inherent in the speech of various audio recordings of speeches and words uttered by dozens of children who took several tests.
The researchers found that the algorithm was effective in diagnosing internalizing disorders in children with 80% accuracy. In addition, the algorithm took only a few seconds to process the piece of speech that was fed to them and make the diagnosis. In particular, the algorithm identified several unique characteristics that stood out for being highly indicative of internalizing disorders and among these there was a voice with a low frequency, with repeated inflections and content in speech.