There is already a mobile application that records a speech, automatically analyzes it and detects, according to speech patterns, the number of verbs used by the speaker, discursive coherence, among other factors, the probability of suffering from schizophrenia.
At the Institute of Research in Computer Science (ICC, CONICET-UBA) Dr. Diego Fernandez Slezak, researcher of the National Council of Scientific and Technical Research (CONICET), is working with this objective, directly related to psychiatry. artificial intelligence and neurosciences.
This application is still in the prototype phase and it is extremely important to obtain data through psychiatrists in order to complete the tool development, to have the possibility for professionals to use the App to conduct patient interviews and, through this data, to be able to do the studies in the laboratory, in order to optimize the prediction capacity, as if this tool was not enough, in addition, it could be used for plagiarism tests or as a lie detector.
This magnificent idea emerged in 2014, when the team led by Fernandez Slezak released a document with the support of the Columbia team, on the speech study of 34 patients who had attended a guard in the United States. They had access to texts of potential schizophrenic patients, they were interviews that the psychiatrists had done to them in the guard. From those interviews, American psychiatrists had not been able to diagnose patients beyond identifying them as "clinically high risk" (CHR), because for the diagnosis of schizophrenia a follow-up of months of interviews and protocols is needed, unless the patient that goes to the guard is in the middle of an outbreak.
After that they developed an automatic analysis of the texts and quantified the messages through certain characteristics, and predicting which high-risk patients were going to unleash schizophrenia, all this reminds them of their main procursor. In the end what was obtained was that of those 34 patients, five suffered completely verified schizophrenia, that is to say, that the calculation of the group through their method was 100% effective.
Obtaining this great experience and these remarkable results, now it is a challenge to perform and validate the exams with more patients and thus obtain the idea in a reliable way in a mobile application that can be used to shorten the times of diagnosis in the practice of psychiatrists.