{"ok":true,"nota":{"id":"8a28a39f4fa25587","slug":"una-computadora-aprendio-a-reconocer-celulas-de-sangre-a-partir-de-dia-8a28a3","titulo":"Una computadora aprendió a reconocer células de sangre a partir de diagnósticos ya hechos","resumen":"Investigadores desarrollaron un sistema de inteligencia artificial que reconstruye cómo se separan las células en estudios de citometría de flujo usando solo los porcentajes finales de cada muestra. No necesitó que alguien marcara célula por célula.","fuente":"medrxiv","fuenteNombre":"medRxiv · preprints","color":"#1f7a4d","autor":"Buser, M., Szczepanowski, M., Laqua, A., Chitadze, G., Baldus, C., Brueggemann, M., Marr, C.","url":"https://www.medrxiv.org/content/10.64898/2026.10.20.26365931v1?rss=1","imagen":"/hoy/portada/8a28a39f4fa25587","tema":"ciencia","subtemas":["inteligencia artificial","citometría de flujo","hematología","diagnóstico médico","aprendizaje automático"],"entidades":["IMPGAT"],"formato":"dato","entretenimiento":false,"pubTs":"2026-10-21T00:00:00.000Z","indexTs":"2026-10-22T04:37:56.220Z","hash":"9ec0cac207b14ee901347e61678fcb1771c543633481bb841ea81db92854f399","tsa":{"time":"Oct 22 04:38:37 2026 GMT","provider":"freetsa.org","sobre":"sello"},"stats":{"views":0,"likes":0,"clicks":0,"dwell":0},"tipo":"normal","registro":"hallazgo","porQueImporta":"Los laboratorios de hematología acumulan millones de análisis de sangre con diagnósticos expertos, pero los datos celda por celda casi nunca quedan guardados de forma usable. Este método permite aprovechar ese archivo para entrenar herramientas de diagnóstico sin rehacer el trabajo manual. Para un paciente argentino, podría traducirse en diagnósticos más consistentes entre laboratorios. Todavía falta probarlo en distintos equipos y poblaciones antes de usarse en la práctica.","quienFirma":"medRxiv (preprint; institución principal no indicada en la fuente)","confianza":"preprint","consulta":null},"sello":{"hash":"9ec0cac207b14ee901347e61678fcb1771c543633481bb841ea81db92854f399","sello":"ff074c3e133af104266a8c1ebac4f2f95fabf928b8cf01a13ef344c96682227b","beacon":{"drand":{"cadena":"8990e7a9aaed2ffed73dbd7092123d6f289930540d7651336225dc172e51b2ce","ronda":6573763,"aleatorio":"ac1ab9766ef209323093f9521e7402b29370fa8f04af200361451518fae9c9b2","tiempo":"2026-10-22T04:38:30.000Z"},"nist":{"cadena":2,"pulso":1994087,"salida":"2CAE0B0247F8B2C1365D5848B4971A88B86F8096596AB659C9B46851CC37B71B43230BE62392778D57CA4D0DAE3440419DC67D4E27A84A89E3AA5B314736F2A3","tiempo":"2026-10-22T04:38:00.000Z"},"noAntes":"2026-10-22T04:38:30.000Z"},"verificacion":{"drand":true,"nist":true},"tsa":{"time":"Oct 22 04:38:37 2026 GMT","provider":"freetsa.org","sobre":"sello"},"original":{"titulo":"Implicit reconstruction of gating strategies from sample-level flow cytometry annotations","descripcion":"Flow cytometry is central to clinical hematology, generating extensive archives comprising billions of single-cell profiles linked to expert-derived diagnostic summaries. While these records represent a largely untapped repository of collective expert knowledge, the underlying single-cell annotations required for supervised machine learning are typically not recorded in a standardized, accessible format. We present IMPGAT (implicit gating), a weakly supervised multiple-instance learning framework that reconstructs single-cell gating strategies directly from existing sample-level cell-type frequency distributions. The model utilizes an induced set attention mechanism to map single-cell marker","pubDate":"2026-10-21","url":"https://www.medrxiv.org/content/10.64898/2026.10.20.26365931v1?rss=1"},"indexado":"2026-10-22T04:37:56.220Z"}}