WOW: Speech Analytics Tool Predicts Schizophrenia With 100% Accuracy
An automated data analytics program can be used to predict psychosis diagnoses in at-risk young people, according to a study published in the journal Schizophrenia, The Atlantic reports (Lafrance, The Atlantic, 8/26).
Details of Study
The study included 34 patients ages 14 to 27 (Bedi et al., Schizophrenia, 8/26).
Researchers used an automated speech analysis program to find "jarring disruptions" in otherwise normal speech. Disorganized thought -- reflected in disjointed speech patterns -- is considered a key trait of schizophrenia.
The researchers' semantic analysis measured coherence and two syntactic markers of speech complexity:
- Sentence length; and
- Number of clauses per sentence.
The study did not review acoustic features that can be meaningful in interpreting speech and thought patterns, such as:
- Intonation; and
The researchers found that the algorithm could distinguish with 100% accuracy which young people would develop psychosis over a two-and-a-half year period and which would not.
Guillermo Cecchi, a biometaphorical-computing researcher for IBM Research and one of the study authors, said, "[W]e found that minimal semantic coherence -- the flow of meaning from one sentence to the next -- was characteristic of those young people at risk who later developed psychosis," which "was not the average." He explained, "As an interviewer, if my mind wandered briefly, I might miss it. But a computer would pick it up."
According to The Atlantic, the program performed better than other advanced screening methods, including:
- Biomarkers from neuroimaging; and
- Electroencephalography recordings of brain activity.
The top predictive models find psychosis developments based on speech tracking with about 79% accuracy.
Researchers plan to try to replicate the findings using transcripts from a larger cohort of at-risk youth (The Atlantic, 8/26).
The Usability People work with you on improving the Usability of Healthcare IT.
Together we may save a life! #SafeHealthIT