{"ok":true,"nota":{"id":"ec32c9e5518c2fcc","slug":"un-algoritmo-analiza-como-habla-la-gente-con-esquizofrenia-y-los-resul-ec32c9","titulo":"Un algoritmo analiza cómo habla la gente con esquizofrenia y los resultados dependen del modelo que se use","resumen":"Compararon grabaciones de habla de 228 personas, 152 con trastornos del espectro esquizofrénico y 76 voluntarios sanos. Probaron seis modelos de lenguaje distintos para medir la similitud semántica entre frases y comprobaron que las conclusiones cambian según la herramienta.","fuente":"medrxiv","fuenteNombre":"medRxiv · preprints","color":"#1f7a4d","autor":"Nikzad, A. H., Cong, Y., Partlan, R., Li, J., Cho, S., Tang, S. X.","url":"https://www.medrxiv.org/content/10.64898/2026.10.23.26366346v1?rss=1","imagen":"/hoy/portada/ec32c9e5518c2fcc","tema":"ciencia","subtemas":["esquizofrenia","inteligencia artificial","salud mental","lenguaje","diagnóstico"],"entidades":["GloVe","Word2Vec","Llama","GPT-2","medRxiv"],"formato":"dato","entretenimiento":false,"pubTs":"2026-10-26T00:00:00.000Z","indexTs":"2026-10-26T16:35:59.047Z","hash":"693b55ffa6e54aa937f54a48ac091fc4c79ac32167f0c7e507e1173e3d70704a","tsa":{"time":"Oct 26 16:36:53 2026 GMT","provider":"freetsa.org","sobre":"sello"},"stats":{"views":0,"likes":0,"clicks":0,"dwell":0},"tipo":"normal","registro":"hallazgo","porQueImporta":"Hay quienes creen que una máquina puede detectar la esquizofrenia escuchando cómo habla una persona, pero este trabajo muestra que el resultado depende de qué programa se use. Con seis modelos distintos, las diferencias entre enfermos y sanos aparecieron en direcciones contradictorias. Para el público general, la lección es que las promesas de diagnóstico con inteligencia artificial necesitan mucha más prueba antes de llegar a un consultorio. El estudio tiene una muestra chica y se centra en un solo tipo de análisis del lenguaje.","quienFirma":"Universidad de Hong Kong / institución psiquiátrica asociada (China)","confianza":"preprint","consulta":null},"sello":{"hash":"693b55ffa6e54aa937f54a48ac091fc4c79ac32167f0c7e507e1173e3d70704a","sello":"8b747273195739447917a7ed6dbe9e9447a039ca94a50678e49748877931aa06","beacon":{"drand":{"cadena":"8990e7a9aaed2ffed73dbd7092123d6f289930540d7651336225dc172e51b2ce","ronda":6586719,"aleatorio":"69c104ecd3cb13434712414db5b2ad3c4f3e8c23a6e958cf0c536314410db9d8","tiempo":"2026-10-26T16:36:30.000Z"},"nist":{"cadena":2,"pulso":2000553,"salida":"C52F940F6CB3ED0E0B156BE5FA3EFA2B75E456473C90354367E414D5115CEDF0D632503093B1A864228DF7BF37C390F188E14CB6127FE2E67FC4E9BF01B2FA0B","tiempo":"2026-10-26T16:36:00.000Z"},"noAntes":"2026-10-26T16:36:30.000Z"},"verificacion":{"drand":true,"nist":true},"tsa":{"time":"Oct 26 16:36:53 2026 GMT","provider":"freetsa.org","sobre":"sello"},"original":{"titulo":"Systematic evaluation of semantic similarity across language models for schizophrenia spectrum disorders","descripcion":"Background: Semantic similarity features derived from language embeddings have been proposed as markers of schizophrenia spectrum disorders (SSD), but prior studies have reported inconsistent directions of effect and limited evidence regarding their relationship to specific dimensions of thought and language disturbance. We systematically evaluated how language model architecture and preprocessing strategy influence semantic similarity signals in SSD. Methods: Speech samples from 228 participants (152 SSD, 76 healthy volunteers) were analyzed using six embedding models: GloVe, LSA, Word2Vec, RoBERTa, GPT-2, and Llama. Semantic similarity features were computed across three preprocessing leve","pubDate":"2026-10-26","url":"https://www.medrxiv.org/content/10.64898/2026.10.23.26366346v1?rss=1"},"indexado":"2026-10-26T16:35:59.047Z"}}