de Florence Rossier
Le Temps /19.09.15
En écoutant la parole des patients, des logiciels sont capables de détecter les signes avant-coureurs de certaines maladies mentales. Une révolution médicale s’annonce
"L’ordinateur parviendra-t-il à surpasser un psychiatre expérimenté pour repérer les personnes à haut risque de schizophrénie ? C’est la surprenante question que pose une étude pilote américaine, publiée le 26 août dans la revue npj Schizophrenia . Une équipe du Centre médical de l’Université Columbia (New York) a évalué les performances d’un logiciel d’analyse de la parole, conçu par IBM, pour dépister les anomalies du langage – l’un des signes de cette maladie psychiatrique" (fin citation)
Pour lire l'article, cliquer ici
► Article original en anglais de "npj Schizophrenia"Automated analysis of free speech predicts psychosis onset in high-risk youths"
Editorial Summary:
Diagnostics: Automated speech analysis predicts later psychosis
A computer program that analyses natural speech could help predict the onset of psychosis in young people at risk. People with schizophrenia have subtle disorganization in speech, even before they first develop psychosis. In a collaboration between IBM, Columbia University Medical Center, and researchers in South America, an automated program that simulates how the human brain understands language was used to analyze interview transcripts from 34 ‘at risk’ youths. Decrease in the flow of meaning from one spoken phrase to the next, and grammatical markers of speech complexity, identified the five individuals who later developed psychosis. The computer program outperformed clinical assessments in predicting psychosis. While numbers are small in this proof-of-principle study, the authors suggest automated analysis could lay the foundation for a simple clinical test of emerging schizophrenia, which would inform early intervention.
► Article sur le site IBM (en anglais): Discriminative Network Models of Schizophrenia
"Schizophrenia is a complex psychiatric disorder that has eluded a characterization in terms of local abnormalities of brain activity, and is hypothesized to affect the collective, “emergent” working of the brain. We propose a novel data-driven approach to capture emergent features using functional brain networks [4] extracted from fMRI data, and demonstrate its advantage over traditional region-of-interest (ROI) and local, task-specific linear activation analyzes. Our results suggest that schizophrenia is indeed associated with disruption of global brain properties related to its functioning as a network, which cannot be explained by alteration of local activation patterns. Moreover, further exploitation of interactions by sparse Markov Random Field classifiers shows clear gain over linear methods, such as Gaussian Naive Bayes and SVM, allowing to reach 86% accuracy (over 50% baseline - random guess)."