Diagnosing depression with the artificial intelligence of the smartphone and the use we make of it, the proposal from Stanford
Depression is a disease that affects more than 300 million people worldwide, of which more than half (more than 90% in many countries) do not receive the necessary treatment, either due to lack of resources economic, diagnostic errors or because of the social stigmatization of mental disorders.
Hence the importance of having an early diagnosis. For this reason, a team of researchers from Stanford University have discovered that the same class of algorithms that allow our mobile devices to offer us facial and voice recognition could in the future allow the option of detecting our first signs of depression with a level reasonable precision.
How to diagnose depression with the use of mobile phones
The scientists turned to a neural network, which they fed with a database (the DAIC-WOZ) that contains video interviews with almost 200 people, both depressed and healthy. In all cases, the interlocutor was an interactive avatar controlled by a doctor.
Each video is represented as a three-dimensional model of a human face during a conversation and a spectrogram of its speech. The objective? Detect patterns and micromanages (facial expressions, tone of voice, fluency of speech, etc.) that could be acting as indicators of this disease.
The results were quite promising: the average error of the AI was 3.67 points, which corresponds to an accuracy of 85% when successfully detecting depression in the study subjects.
And while this research is still at an early stage, the researchers believe that one day "this technology could be implemented on any smartphone and thus facilitate low-cost universal access to mental care."
Although they warn, yes, that such technology would not replace, in any case, the work of medical professionals.