{"ok":true,"nota":{"id":"67770e50b7513207","slug":"una-herramienta-de-ia-para-el-arn-no-codificante-mejora-poco-frente-a--67770e","titulo":"Una herramienta de IA para el ARN no codificante mejora poco frente a los métodos de siempre","resumen":"Investigadores compararon modelos de inteligencia artificial que predicen la actividad de ARN no codificante contra métodos tradicionales. Midieron cuánto aporta cada componente sobre una referencia base, con datos de experimentos en distintas células.","fuente":"biorxiv","fuenteNombre":"bioRxiv · preprints","color":"#be2628","autor":"Guo, R., Shan, W., Liu, R., Yang, L., Mu, D., WANG, H.","url":"https://www.biorxiv.org/content/10.64898/2026.10.22.760090v1?rss=1","imagen":"/hoy/portada/67770e50b7513207","tema":"ciencia","subtemas":["inteligencia artificial","genética","arn no codificante","biología computacional","evaluación de modelos"],"entidades":["AlphaGenome","bioRxiv"],"formato":"dato","entretenimiento":false,"pubTs":"2026-10-25T00:00:00.000Z","indexTs":"2026-10-26T04:38:59.737Z","hash":"07de50e3ca3fd04f8ae1f5d3b2bb2e088eab771e6e032b7f8204cd8fc55f1942","tsa":{"time":"Oct 26 04:39:51 2026 GMT","provider":"freetsa.org","sobre":"sello"},"stats":{"views":1,"likes":0,"clicks":0,"dwell":5},"tipo":"motor","registro":"hallazgo","porQueImporta":"En algunos casos la IA sumó bastante, por ejemplo al pasar la precisión de 0,64 a 0,75 en un tipo de experimento. En otros el aporte fue casi nulo o negativo, lo que muestra que no alcanza con que el modelo prediga bien para saber si realmente agrega algo. Para quien sigue las promesas de la IA en medicina, esto obliga a pedir la comparación con lo que ya existía antes de creer en el titular. Conviene desconfiar del entusiasmo que no muestra su base.","quienFirma":"bioRxiv (preprint, instituciones no especificadas en la fuente)","confianza":"preprint","consulta":{"pregunta":"¿Confiarías en una IA médica sin ver su comparación con lo ya existente?"}},"sello":{"hash":"07de50e3ca3fd04f8ae1f5d3b2bb2e088eab771e6e032b7f8204cd8fc55f1942","sello":"90e7af2e9cbfb3adddb835ec23c51e09cdb5194f1861dc417173e7080b0da644","beacon":{"drand":{"cadena":"8990e7a9aaed2ffed73dbd7092123d6f289930540d7651336225dc172e51b2ce","ronda":6585285,"aleatorio":"eb615b82e2756a5239e083f84b3ee4cca3905aac866b8093e58c2109a1b6ba65","tiempo":"2026-10-26T04:39:30.000Z"},"nist":{"cadena":2,"pulso":1999836,"salida":"8A9648E112C884754C4BA14CEABA6D2AFA1376B135546D6B47CC74503EB82E37828C3076A34D73719D10498F27CA2A2BF907EE9977E40FD1B83A29275B0FFCE0","tiempo":"2026-10-26T04:39:00.000Z"},"noAntes":"2026-10-26T04:39:30.000Z"},"verificacion":{"drand":true,"nist":true},"tsa":{"time":"Oct 26 04:39:51 2026 GMT","provider":"freetsa.org","sobre":"sello"},"original":{"titulo":"Evaluating the added value of sequence models in noncoding RNA assays: a baseline-aware framework","descripcion":"Sequence models are evaluated against non-coding RNA assay read-outs, but prediction alone does not show what their features add to an explicit baseline. We estimated model-added-last changes in AUROC for lncRNA labels and cross-validated Spearman correlation for miRNA repression, fixing members and chromosome folds across available sources. In cross-cell CRISPRi (n = 2961), AlphaGenome increased AUROC from 0.6426 to 0.7551; adding K562 accessibility changed these values to 0.7282 and 0.7534. Corrected-source Cas13 increments ranged from -0.0161 to +0.0130 (n = 871). Separate post-audit comparisons tested fixed sequence composition, broader regularization and TargetScan aggregate identity. W","pubDate":"2026-10-25","url":"https://www.biorxiv.org/content/10.64898/2026.10.22.760090v1?rss=1"},"indexado":"2026-10-26T04:38:59.737Z"}}