Centre d'Information et de documentation du CRA Rhône-Alpes
CRA
Informations pratiques
-
Adresse
Centre d'information et de documentation
du CRA Rhône-Alpes
Centre Hospitalier le Vinatier
bât 211
95, Bd Pinel
69678 Bron CedexHoraires
Lundi au Vendredi
9h00-12h00 13h30-16h00Contact
Tél: +33(0)4 37 91 54 65
Mail
Fax: +33(0)4 37 91 54 37
-
Détail de l'auteur
Auteur Arndis SIMONSEN |
Documents disponibles écrits par cet auteur (1)
Faire une suggestion Affiner la recherche
Vocal markers of autism: Assessing the generalizability of machine learning models / Astrid RYBNER in Autism Research, 15-6 (June 2022)
[article]
Titre : Vocal markers of autism: Assessing the generalizability of machine learning models Type de document : Texte imprimé et/ou numérique Auteurs : Astrid RYBNER, Auteur ; Emil TRENCKNER JESSEN, Auteur ; Marie DAMSGAARD MORTENSEN, Auteur ; Stine Nyhus LARSEN, Auteur ; Ruth GROSSMAN, Auteur ; Niels BILENBERG, Auteur ; Cathriona CANTIO, Auteur ; Jens Richardt MØLLEGAARD JEPSEN, Auteur ; Ethan WEED, Auteur ; Arndis SIMONSEN, Auteur ; Riccardo FUSAROLI, Auteur Article en page(s) : p.1018-1030 Langues : Anglais (eng) Mots-clés : Autism Spectrum Disorder Autistic Disorder/diagnosis Biomarkers Humans Machine Learning Speech Voice biobehavioral markers generalizability Index. décimale : PER Périodiques Résumé : Machine learning (ML) approaches show increasing promise in their ability to identify vocal markers of autism. Nonetheless, it is unclear to what extent such markers generalize to new speech samples collected, for example, using a different speech task or in a different language. In this paper, we systematically assess the generalizability of ML findings across a variety of contexts. We train promising published ML models of vocal markers of autism on novel cross-linguistic datasets following a rigorous pipeline to minimize overfitting, including cross-validated training and ensemble models. We test the generalizability of the models by testing them on (i) different participants from the same study, performing the same task; (ii) the same participants, performing a different (but similar) task; (iii) a different study with participants speaking a different language, performing the same type of task. While model performance is similar to previously published findings when trained and tested on data from the same study (out-of-sample performance), there is considerable variance between studies. Crucially, the models do not generalize well to different, though similar, tasks and not at all to new languages. The ML pipeline is openly shared. Generalizability of ML models of vocal markers of autism is an issue. We outline three recommendations for strategies researchers could take to be more explicit about generalizability and improve it in future studies. LAY SUMMARY: Machine learning approaches promise to be able to identify autism from voice only. These models underestimate how diverse the contexts in which we speak are, how diverse the languages used are and how diverse autistic voices are. Machine learning approaches need to be more careful in defining their limits and generalizability. En ligne : http://dx.doi.org/10.1002/aur.2721 Permalink : https://www.cra-rhone-alpes.org/cid/opac_css/index.php?lvl=notice_display&id=476
in Autism Research > 15-6 (June 2022) . - p.1018-1030[article] Vocal markers of autism: Assessing the generalizability of machine learning models [Texte imprimé et/ou numérique] / Astrid RYBNER, Auteur ; Emil TRENCKNER JESSEN, Auteur ; Marie DAMSGAARD MORTENSEN, Auteur ; Stine Nyhus LARSEN, Auteur ; Ruth GROSSMAN, Auteur ; Niels BILENBERG, Auteur ; Cathriona CANTIO, Auteur ; Jens Richardt MØLLEGAARD JEPSEN, Auteur ; Ethan WEED, Auteur ; Arndis SIMONSEN, Auteur ; Riccardo FUSAROLI, Auteur . - p.1018-1030.
Langues : Anglais (eng)
in Autism Research > 15-6 (June 2022) . - p.1018-1030
Mots-clés : Autism Spectrum Disorder Autistic Disorder/diagnosis Biomarkers Humans Machine Learning Speech Voice biobehavioral markers generalizability Index. décimale : PER Périodiques Résumé : Machine learning (ML) approaches show increasing promise in their ability to identify vocal markers of autism. Nonetheless, it is unclear to what extent such markers generalize to new speech samples collected, for example, using a different speech task or in a different language. In this paper, we systematically assess the generalizability of ML findings across a variety of contexts. We train promising published ML models of vocal markers of autism on novel cross-linguistic datasets following a rigorous pipeline to minimize overfitting, including cross-validated training and ensemble models. We test the generalizability of the models by testing them on (i) different participants from the same study, performing the same task; (ii) the same participants, performing a different (but similar) task; (iii) a different study with participants speaking a different language, performing the same type of task. While model performance is similar to previously published findings when trained and tested on data from the same study (out-of-sample performance), there is considerable variance between studies. Crucially, the models do not generalize well to different, though similar, tasks and not at all to new languages. The ML pipeline is openly shared. Generalizability of ML models of vocal markers of autism is an issue. We outline three recommendations for strategies researchers could take to be more explicit about generalizability and improve it in future studies. LAY SUMMARY: Machine learning approaches promise to be able to identify autism from voice only. These models underestimate how diverse the contexts in which we speak are, how diverse the languages used are and how diverse autistic voices are. Machine learning approaches need to be more careful in defining their limits and generalizability. En ligne : http://dx.doi.org/10.1002/aur.2721 Permalink : https://www.cra-rhone-alpes.org/cid/opac_css/index.php?lvl=notice_display&id=476