2023 Author: Bryan Walter | [email protected]. Last modified: 2023-05-21 22:24
American developers have learned to predict the development of psychosis in a patient. To do this, they used an automatic speech analysis algorithm, which included two metrics: an assessment of the semantic content of speech and the use of words associated with sounds. It turned out that those patients who developed psychosis shortly after the interview were characterized by poor vocabulary and frequent use of words like "sound" and "whisper." The accuracy of such a diagnosis, as reported in an article published in npj Schizophrenia, was 93 percent.
Many mental disorders (including, for example, schizophrenia and bipolar disorder) are accompanied by psychosis - an altered state of the psyche, in which connection with the objective external world is lost, and a person begins to see, hear and feel something that is not actually happening. As such, there is no treatment for psychosis: depending on the accompanying diagnosis, antipsychotic drugs or therapy may help, but they rarely manage to get rid of the condition permanently with the help of them.
In this case, psychosis does not necessarily arise as a result of the development of some kind of disease and is included in its pathogenesis, or it may even be a primary syndrome. Mild psychosis can be controlled, due to which it is possible to get rid of the subsequent deterioration of the mental state. To facilitate the diagnosis of psychosis, automatic methods are now being actively developed: many of them are aimed at determining the likelihood of developing psychosis in a person's speech, which changes due to a change in mental state. In January last year, developers, for example, were able to use automatic speech analysis to identify a number of factors signaling the likely development of psychosis (a violation of semantic coherence, a decrease in the length of sentences, the absence of demonstrative pronouns) and achieve a prediction accuracy of 83 percent.
A new similar method was proposed by scientists led by Phillip Wolff from Emory University (Atlanta, USA). Unlike their predecessors, they focused on semantics, or rather, on two aspects of human speech that are responsible for the meaning of what is said and can signal the risk of developing psychosis.
The first aspect is the semantic poverty of speech: the presence of obsessive thoughts, lack of connection with reality can lead to the fact that a person's speech becomes impoverished, the vocabulary used greatly decreases, and certain repetitive patterns appear. The second aspect is directly related to one of the most striking symptoms of psychosis - auditory hallucinations: their appearance, of course, does not go unnoticed, which is why a person experiencing them may more often talk about what he hears using words associated with sounds and voices.
In their study, the researchers used clippings from interviews of 40 people who were observed by a university psychiatrist for two years: of these, 12 people were diagnosed with psychosis during the observation (and after the interview). The researchers used the word2vec algorithm, which represents each word as a vector in a multidimensional space; in it, the proximity of two vectors to each other indicates the semantic proximity of the two words.
In order to determine a certain field of the semantic norm of ordinary conversations and, accordingly, to pre-train the model, scientists have compiled a vector representation of words from posts on Reddit of 30 thousand users. After the vector representation of each word, the algorithm composes a vector representation of each sentence, summing the values of the vectors denoting the words in it, on the basis of which the main idea-sentence is allocated, which is also represented as a vector corresponding to a certain word in the sentence. For example, the unifying and basic meaning of the sentences “I want an apple”, “he wants to live” and “they want freedom” is the meaning of the need for something, which is expressed in the verb “want”. The semantic content of the sentences used is calculated by calculating the proximity of the main word in it to all others. The same model was used to highlight the semantic field of words associated with sounds: it included, for example, the words "whisper", "hear", "loud", "voice" and "sound".
On the basis of both models, a classifier was then trained, which, based on the indicator of semantic fullness and the presence of words associated with sounds, determines whether a person falls into a group of people who develop psychosis. It turned out that both indicators correlate with whether a person will be diagnosed with psychosis. A combined analysis of patient interviews using the two metrics made it possible to correctly predict the onset of psychosis with an accuracy of 93 percent.
Despite the fact that the used algorithm works on the basis of speech analysis - just like diagnostics by a psychiatrist - the new method can be much more effective: an objective statistical analysis of the semantic parameters of speech is based on working with more material. This also means that two separate possible markers of the development of psychosis - the use of words associated with sounds and the poverty of vocabulary - in psychotherapeutic practice cannot be used to make a diagnosis.
Psychosis is not the only thing that people now know how to diagnose with the help of automatic speech analysis. In the fall, American developers, for example, presented an algorithm that, based on speech analysis, determines depression with an accuracy of 77 percent.
