It was created by Chilean researchers. It is based on biomarkers of spoken language. Thanks to that, the algorithm that predicts schizophrenia was developed. A tool with great potential for the diagnosis of this mental illness.
The Faculty of Medicine of the University of Chile promoted the project. It was led by Alicia Figueroa Barra, a specialist in clinical linguistics applied to psychiatry and the study of schizophrenia.
An automated language analysis was performed on 133 Spanish-speaking study subjects. 49 healthy; 40 with a first psychotic break and 44 persons with chronic schizophrenia. They detected 30 features of spoken language, grouped into three variables.
Verbal fluency: takes into account the continuity of speech, including pauses and hesitations of the speaker. Thus, people with a psychotic break had longer speech pauses.
Verbal productivity: this refers to the ability to pronounce a number of words and sentences. Poverty of speech” is characteristic of the language of people with schizophrenia.
Semantic coherence: is the logical organization of meaning in speech. Patients with schizophrenia often change topics suddenly during conversation. That is, they use wrong words that affect the referentiality and concordance of what they say.
“We trained an automatic classification algorithm. Coherence is the most relevant feature,” explained computer engineer Mauricio Cerda. He is one of the researchers in the study.
Combined with other disease diagnostic tools, such as demographic data and PANSS test responses. They are used in the diagnosis of schizophrenia. An accuracy of 77.5% was achieved in predicting whether a patient would have schizophrenia.
Therefore, it is concrete, measurable and quantifiable information. Because of this, the algorithm that predicts schizophrenia is a clear supportive tool. It would improve early detection of the disease.
“So, thanks to language analysis we can save costly interventions for the patient. Emotionally and economically,” said linguist Figueroa Barra.