{"ok":true,"nota":{"id":"60adf32fa505d219","slug":"un-modelo-de-ia-mas-barato-y-rapido-ayuda-a-clasificar-el-riesgo-quiru-60adf3","titulo":"Un modelo de IA más barato y rápido ayuda a clasificar el riesgo quirúrgico, con pocos datos locales","resumen":"Investigadores compararon varias formas de usar un modelo de lenguaje grande para clasificar el estado físico de pacientes antes de una cirugía, según la escala de la Sociedad Americana de Anestesiólogos. Midieron precisión, velocidad, costo y cuántos casos etiquetados por humanos necesitaba cada método.","fuente":"medrxiv","fuenteNombre":"medRxiv · preprints","color":"#1f7a4d","autor":"Shi, Y., Wang, Y., Simayi, A., Qu, L., Xu, G.","url":"https://www.medrxiv.org/content/10.64898/2026.10.20.26366028v1?rss=1","imagen":"/hoy/portada/60adf32fa505d219","tema":"ciencia","subtemas":["inteligencia artificial","anestesiología","clasificación clínica","riesgo quirúrgico","modelos de lenguaje"],"entidades":["Sociedad Americana de Anestesiólogos","medRxiv"],"formato":"dato","entretenimiento":false,"pubTs":"2026-10-22T00:00:00.000Z","indexTs":"2026-10-22T22:02:10.729Z","hash":"4e7b07a79ffc1175be9e83e0a7f70ac546e0efa20f52ab4c3cd94eea6ba23e5d","tsa":{"time":"Oct 22 22:02:55 2026 GMT","provider":"freetsa.org","sobre":"sello"},"stats":{"views":0,"likes":0,"clicks":0,"dwell":0},"tipo":"motor","registro":"hallazgo","porQueImporta":"Un hospital del interior que no tiene miles de casos clasificados por anestesiólogos podría adoptar una herramienta así con apenas unas decenas de ejemplos propios. Eso baja costos y tiempos en un paso que hoy depende del criterio de cada profesional. Todavía falta confirmar que funciona igual de bien en hospitales distintos, y la decisión final sigue siendo del médico.","quienFirma":"medRxiv (preprint; institución principal no identificada en el resumen)","confianza":"preprint","consulta":{"pregunta":"¿Confiarías en una IA para ayudar a evaluar tu riesgo antes de una operación?"}},"sello":{"hash":"4e7b07a79ffc1175be9e83e0a7f70ac546e0efa20f52ab4c3cd94eea6ba23e5d","sello":"3343515ef502555b28696c70ec495db130c40c8a7c2239b02201a05422fd9b4a","beacon":{"drand":{"cadena":"8990e7a9aaed2ffed73dbd7092123d6f289930540d7651336225dc172e51b2ce","ronda":6575851,"aleatorio":"29e6764b49393fcb685b41ab04eb6667b179de6b518c8212d4aa11d1de9d1b19","tiempo":"2026-10-22T22:02:30.000Z"},"nist":{"cadena":2,"pulso":1995125,"salida":"70391569C3CDB488E71ABC660C5481B4A24B300E2E08E833696854B6488DAB0370809D6AA479035A0F6218DCEDD92C536F46C33AF62EF2BFF2B7635A2FD7E526","tiempo":"2026-10-22T22:02:00.000Z"},"noAntes":"2026-10-22T22:02:30.000Z"},"verificacion":{"drand":true,"nist":true},"tsa":{"time":"Oct 22 22:02:55 2026 GMT","provider":"freetsa.org","sobre":"sello"},"original":{"titulo":"Non-generative readout with few-label calibration versus LoRA fine-tuning of a large language model for ASA physical status classification: accuracy, speed, cost and label efficiency in three data sources","descripcion":"Background. The American Society of Anesthesiologists Physical Status (ASA-PS) class underpins perioperative risk communication, but its assignment is subjective. Generative large language models (LLMs) are slow and costly, and LoRA fine-tuning needs site-specific labels and a separately maintained adapter for every site. Non-generative decision models return option probabilities in one forward pass, but their comparison with LoRA in anesthesia, and the number of local labels they need, is unstudied. Objective. To compare, on one open base model, generative zero-shot prompting, a non-generative readout, the readout with cutpoint calibration on a few local labels, and task LoRA for four-class","pubDate":"2026-10-22","url":"https://www.medrxiv.org/content/10.64898/2026.10.20.26366028v1?rss=1"},"indexado":"2026-10-22T22:02:10.729Z"}}